This is the conversation about proof. Specifically, financial proof.

The single biggest change from an early-stage room to a growth-stage one isn't a folder. It's what happens inside 02_financials. At the earliest stages, your model is simply a basic projection and your unit economics are things like CAC and LTV. At the later stages, the bar is exponentially higher and the financial narrative has to arguably be the most robust component of your entire fundraising infrastructure. Your financials need to be prepared by an actual finance team and should be polished, thorough, and buttoned-up. Ideally, your financial narrative needs to survive the rigors of public market scrutiny or an investment banking due diligence process.

Before I dig into the details, I feel it’s important to share that I don’t consider myself an expert on finance. I almost failed both Managerial Accounting and Financial Management class in business school. And, like many of you, I feel overwhelmed and intimidated when I open an expansive financial model with a dozen tabs all linked to one another. But as the former Head of Investor Relations at GrubMarket, I was forced to figure out how to navigate my way around pre-IPO level financials quickly, and not only build models that passed the most rigorous Tier-1 banking standards, but also be able to explain the model line-by-line and defend every single number shown in an Excel file. This was easily the most challenging part of any job I’ve ever had. But with the help of some financially savvy colleagues and an intense ‘figure it out’ mentality, I was able to overcome my crippling fear of finance. I can laugh at the irony of my career now.

This is the primer I wish I had when I was learning the financial ropes with my feet against the fire. I hope you find this helpful. If I can learn how to get comfortable with financials, so can you. Good luck!

Subscribe now

Let’s get started. In this post, I will attempt to teach you how to read financials, and explain key terms and formulas in as simple of a manner as possible. Bear with me.

What’s driving the model

Most people see a financial model as a spreadsheet of outputs. But at its core, it’s really a chain of assumptions that map out to said outputs. A sophisticated investor will focus less on the top-line revenue and spend more time on the underlying handful of inputs that produced it. Broken assumptions = bad outputs, so keep that in mind.

Here are a few of the more common assumptions that drive the numbers you see on the model spreadsheet:

• Pipeline and conversion. The trend you’re looking to understand here is what fraction of the customer pipeline actually becomes live paying customers, and how that rate has changed over the last four quarters. A model that assumes (with no historical data to support it) a constantly flat or consistently improving conversion rate in perpetuity, is an instant red flag. Be reasonable.

• Sales cycle length. This one is more self explanatory. The data here shows how long a deal actually takes to close. Companies often like to show compressing sales cycle lengths over time in order to justify aggressive growth targets. But if you do this with no evidence that your sales motions have actually gotten faster, that’s just sloppy and investors will pick up on it right away.

• Average contract value (ACV) and pricing. ACV trends show whether your company is growing because of better pricing power and stronger upsell performance, or whether you’re showing a mix shift towards a handful of large, one-off deals that aren’t repeatable (or just anomalies).

• Expansion and churn. The NRR and GRR numbers the model assumes going forward should roughly match what the historical waterfall (explained in more detail below) has actually shown. Don’t create unsubstantiated churn improvement numbers, or else your model will lose credibility quickly.

• Headcount and ramp time. Hiring plans and new hire productivity assumptions drive both sides (cost and growth) of the model simultaneously, since un-ramped headcount spends money before it produces revenue. It’s easy to be optimistic with ramp up times, which in turn will overstate near-term growth in the model.

• Cost structure. This one boils down to whether cost of goods sold (or “cost of revenue”) improves, as a percent of revenue, as the company scales, and whether OpEx growth is assumed to slow down compared to revenue growth. This is the operating leverage assumption, and it’s typically the single biggest driver of when, or whether, the model ever shows bottom-line profitability.

The three key statements (IS, BS, CF)

Once the assumptions are set, they flow into three linked statements. Each one answers a different set of questions, and tells the deeper story behind the company’s performance and how it’s run.

At the late-stages, your entire financial statement should ideally be audited, and not just presented as management-prepared numbers. Audited financials means someone independent has actually checked the story against the numbers and verified the raw numbers driving your business performance claims. It’s hard for investors to just inherently trust the numbers presented to them, when management teams are incentivized to put their best foot forward and present only the best (or ideal) version of themselves.

1. The P&L (Profit & Loss statement, sometimes known as the Income Statement)

What it answers: are we making money on what we sell, and how much of what’s left gets re-invested (i.e. spent) on growth.

What this typically looks like, top to bottom:

*Note: EBITDA vs. Adjusted EBITDA. EBITDA strips out financing and accounting effects to show core operating profitability. Sometimes companies will just show *Adjusted* EBITDA, which goes an extra step further and adds back one-time or non-cash items, most commonly things like stock-based compensation, one-time legal fees, and non-recurring restructuring costs. The EBITDA bridge is the itemized waterfall that starts from reported earnings and trickles down to Adjusted EBITDA, showing every add-back individually instead of just netting them into a single number. A clean, defensible bridge builds trust and credibility.

The real nuance lives inside the Operating Expenses section:

• Sales & Marketing (S&M). This section shows the customer acquisition cost. A rising S&M isn’t a problem by itself, but a rising S&M growing faster than the new revenue it produces triggers warning bells. Best practices dictate that you watch S&M against new bookings or new ARR generated in the same period, not S&M as a flat percent of total revenue (which is commonly done, but is more often than not a more flattering depiction).

• R&D / Engineering. The nuance most founders miss here is that some software development cost gets capitalized onto the balance sheet instead of expensed on the P&L, depending on accounting policies. Two companies with identical engineering spend can report vastly different R&D numbers and margins, purely based on how much they capitalize. Comparing R&D-to-revenue across companies can be difficult for this reason.

• G&A (General & Administrative). This represents the back-office cost of running the company: finance, legal, HR, and executive compensation. This line should grow slower than revenue as the company scales. G&A growing faster than revenue usually signals organizational bloat, not growth investment, and it’s one of the first places a diligence process goes looking when a burn multiple looks worse than the growth rate should produce.

• Fixed versus variable cost. Not every OpEx line behaves the same way if growth slows. Things like payment processing fees and sales commissions flex down with revenue, hence variable. Line items such as office leases, base salaries, and software licenses typically don’t, hence fixed. That mix determines how fast a company can actually cut its way to a longer runway.

• One-time versus recurring cost. A cost is only legitimately one-time if it doesn’t show up again next quarter. When the same category is recurring and just relabeled each time, investors will catch on quickly, and start discounting the efficacy of the EBITDA bridge.

Signals to read off of the P&L: gross margin trending upwards usually means better pricing power or a favorable mix shift. Trending downwards typically means rising infrastructure or support costs are eating into the profits. In theory, operating expense as a percent of revenue should decline over time at a scaling company. That’s the signal of true operating leverage.

2. The Balance Sheet

What it answers: what the company owns, what it owes, and what’s left over, at a single point in time (monthly, yearly, etc.).

The accounting equation that always has to hold:

Assets

• Current: cash, accounts receivable, prepaid expenses.

• Non-current: capitalized software, equipment, goodwill and intangibles (if there’s been an acquisition).

Liabilities

• Current: accounts payable, accrued expenses, and, for a subscription business, deferred revenue.

• Non-current: long-term debt, venture debt, convertible notes.

Equity

• Paid-in capital from every round raised, net of the accumulated deficit most startups carry as normal. Not alarming on its own, as long as the trend is narrowing over time rather than widening.

Signals to read off the balance sheet: deferred revenue growing faster than recognized revenue is a genuinely good sign for a subscription-based business. It means that cash is coming in ahead of when it gets counted as revenue. Accounts receivable growing faster than revenue is the opposite signal, and is usually a sign that collections are slipping (worth checking against days sales outstanding). And for anyone carrying venture debt, the covenants attached to it belong in this conversation too.

3. The Cash Flow Statement

What it answers: whether the profit or loss on the P&L actually showed up as real cash in the bank.

Reconciled from net income:

What each section is really telling you:

• Operating activities. This is the cash amount the actual business generates or consumes. If this number turns positive, it signals a big inflection point that the core business, without any financing help, is able to fund itself. Every company should be aiming to become operating cash flow positive.

• Investing activities. This is comprised mostly of CapEx and acquisitions. Heavy spend in this category isn’t automatically a red flag, but it’s worth delineating between whether it’s maintenance (keeping the lights on) or growth (building new capacity), since those require very different justifications in an investor conversation.

• Financing activities. This section outlines the money raised, debt drawn or repaid. What to watch for here: a cash balance growing only because of financing activity, while operating cash flow stays flat or negative. That signals a company that potentially relies too much on inorganic revenue (or rented growth). It’s not a disqualifying factor at the growth stage, but will raise many questions about the longer-term sustainability of the business model or growth compounder.

The metrics that come out of these three statements

Everything below is a derivative of the three key financial statements above. Each one tells a different part of the full financial health story.

Net working capital. Typically calculated as current assets minus current liabilities (mostly receivables, inventory, and payables). A company can grow revenue at an impressive pace, while quietly needing more and more cash just to fund the gap between paying suppliers and collecting from customers. That gap is what a standalone net working capital analysis tracks over time.

The free cash flow waterfall, the same steps as the cash flow statement above, in slightly simpler terms:

The ARR waterfall. A marker of how ARR actually changed this quarter:

This waterfall is important because it reveals the whole story of how you’re growing, not just simply that you are growing.

Net revenue retention and gross revenue retention. These have the same underlying waterfall, but answer two different questions. New revenue from brand-new logos is deliberately excluded from both, since both are only measuring what happened with the existing customer base:

GRR excludes upsell revenue entirely, can never exceed 100%, and tells you squarely how leaky the bucket is. NRR includes expansion revenue and can exceed 100%, which would mean that existing customers are worth more to you over time without needing to onboard a single new logo. Best-in-class software businesses normally show NRR of 120%+. Strong NRR with a weak GRR means the survivors are expanding fast enough to mask a lot of churn underneath, which is a more fragile story, and a trend worth keeping a closer eye on.

Logo retention (by business model). This shows the number of customers retained at the end of a given closing period, divided by the number of customers at the start of the period. This is run quarterly or annually depending on sales cycle. Revenue retention can look fine while a company is quietly losing customers and getting bailed out by the few who have a larger expansion delta, which is exactly why a sophisticated investor asks for both numbers, not just one:

• B2B: 90%+ annually is solid. Losing an enterprise customer means losing a longer-term relationship and a more significant source of revenue.

• B2C: meaningfully lower than B2B almost by design, since switching costs less for an individual than for an enterprise.

• SaaS: tends to report logo retention alongside dollar retention on purpose, because a company can post excellent NRR off of a handful of expanding accounts while quietly bleeding logos underneath. That’s a business that looks healthy but is flashing warning signals.

Burn multiple

The fastest way for a growth investor to sanity-check capital efficiency. Under 1x is elite. Above 2x or 3x, and an investor begins asking whether the growth is worth what it costs to buy.

Rule of 40

A company growing 60% while burning at negative-20% margin passes (60 - 20 = 40), and so does one growing 20% at 20% margin. This metric catches both failure modes at once: growing too slowly to justify the burn, or burning too much to justify the growth.

How it all comes together, stage by stage

Pre-seed to Series A. Investors are buying a story.

The model is a projection; unit economics are things like CAC, LTV, and payback periods; the cap table is simple enough to fit onto one page. The data room itself lives in a Google Drive. Its job is mostly to confirm the story in the deck isn’t contradicted by anything in the paperwork.

Series B/C. Investors are buying a repeatable process.

They are checking the math. That means the model needs real historical actuals to stress-test against. Adjusted EBITDA, burn multiple, and a real FCF waterfall enter the conversation. Unit economics stop being a nice-to-have slide and become the file a growth investor reads first. NRR becomes a headline number. The cap table gets genuinely more complex, and the conversation about a dedicated VDR platform begins.

Series D and beyond. Investors are buying an outcome.

And they are underwriting it like one. At this point, the financials have to hold up to something close to audited scrutiny. Metrics become the same vocabulary that public markets use: NRR, rule of 40, a clean ARR waterfall, an Adjusted EBITDA bridge. The legal, commercial, and tax folders that were nearly empty at pre-seed are now fleshed out and up-to-date. The data room isn’t proving that a story is just plausible anymore. It’s holding up to the same scrutiny a public company gets.

The thread that ties all three “Inside the Data Room” articles together

Whether it’s five folders in a Google Drive or fifteen in Intralinks, every data room is answering the same underlying question, just with rising stakes: is the company in front of me the same company on paper? At pre-seed, “on paper” means does the story hold up. At Series D, it means does this survive the kind of scrutiny a public offering would bring. The question never changes. The bar just keeps rising.

I know I’ve said this before, but build the room like someone smarter than you is already inside it, because by the time it matters, they are. A good pitch deck earns you a meeting. A great data room earns you the wire. I hope you found value in this 3-part series digging into the gnarly nuances of data rooms. Best of luck to all of you founders and aspiring entrepreneurs out there!

Share The Data Room

Last week I walked through the core of a data room: the five folders every company needs before diligence starts, built with early-stage founders in mind. If you haven’t read Part I, start here, because everything in this article sits on top of that foundation. You shouldn’t try to build the advanced folders until the core ones are clean.

This is what the room grows into. Quick housekeeping: I planned this as a two-part series. Once I actually sat down to write out everything that changes at growth stage, folder outline in one part and financial breakdown in another became the clearer path forward, not just one long piece pretending it wasn’t two different conversations. So this is now three parts. This one’s the folders. Next week is a breakdown of the numbers (aka financial due diligence).

Here’s the thing most first-time founders don’t see coming: a data room isn’t a static checklist you build once. It’s a living thing that has to grow up on the same schedule the company does.

The five core folders stay (deck, financials, cap table, legal, metrics), but what goes inside each one gets deeper and several new folders appear. That shift happens because the person on the other side of the table changes. A pre-seed investor is buying a story tested lightly against reality. A Series D investor is buying an outcome, and they’re underwriting it like one, closer to how a public-market analyst would.

Let me show you what gets added, and why.

Subscribe now

06_product.

The core financials tell an investor whether the business works. This folder tells them whether the product is real and where the build vision is going. A roadmap needs to show genuine prioritization logic, not just a wish list of every idea you’ve ever had. Investors are looking to see how you decide what to build.

A technical architecture overview is important if there’s real technical differentiation worth understanding. And be prepared to furnish product engagement metrics that go a level deeper than the top-line KPI dashboard from Part I: activation rate, yes, but also feature adoption, time-to-value, and DAU/MAU ratio (daily active users divided by monthly active users, a measure of how “sticky” the product is). A ratio near 0.5 means half your monthly users show up daily, which is exceptional; near 0.1 means people sign up and drift away.

One more addition that earns its place by growth stage (though you should consider having this in an early-stage data room anyhow): an actual product demo video or recorded walkthrough. A partner who wasn’t in your live demo, or an associate building the investment memo at 11pm the night before a partner meeting, will watch three minutes of screen recording before they’ll schedule a follow-up call to see the product live. It’s a low-effort file that removes a real friction point in how deals actually get championed inside a fund. Early on this folder can be a one-pager. By Series B, “here’s the roadmap” needs to come with “and here’s how last year’s roadmap tracked to what we actually shipped,” because investors have learned to discount roadmaps that always slip or fall behind schedule.

07_market.

This is different from the TAM/SAM/SOM slide in your deck: it’s the underlying research that particular slide is based on. This folder should include items like third-party market sizing reports (if you can access them), a real competitive landscape breakdown (not just a 2x2 chart where you’re conveniently in the top-right corner), customer or user research (surveys, interview notes), and win/loss analysis explaining why you win deals and, more revealingly, why you lose them. But I also want to point out two things I often see growth-stage founders consistently under-build or misunderstand:

08_commercial.

This is the folder that most often lags a company’s actual revenue, and the gap gets dangerous. Items I expect to see here include customer contracts (MSAs, or Master Service Agreements, order forms, whatever your actual paper looks like), a real customer list with logos and reference customers investors can call, and, once there’s a real sales motion, the deal-level metrics a top-line KPI dashboard doesn’t capture:

• ACV (Average Contract Value): the average annualized value of a customer contract. Rising ACV over time is one of the cleanest signals that you’re moving upmarket or successfully expanding accounts.

• Win rate: the percentage of qualified opportunities you actually close. It tells an investor whether your growth is efficient or whether you’re just throwing enormous top-of-funnel volume at a leaky process.

• Sales cycle length: how long a deal takes from first contact to signed. It drives how predictable your growth is and how much cash you have to float before revenue shows up.

• Pipeline coverage ratio: how many dollars of pipeline you have relative to the quota you need to hit; 3x to 4x coverage is a common healthy benchmark. Thin coverage means next quarter is a coin flip.

• Customer concentration: what percentage of revenue sits with your top 3-5 accounts. A “great” $10M revenue number where one customer is $6M of it is a completely different risk profile than the same $10M spread across fifty accounts, and an investor will find that out, so you’re better off surfacing it yourself.

The file most founders skip entirely, but is essential: real customer case studies that are individual account deep-dives, and not just a generic logo wall. These should address what problem this specific customer had and what changed after they bought your product or solution (aka the ROI). For acquisitive businesses (whether it’s a compounding roll up strategy or one-off), this is also where post-acquisition synergy tracking belongs: the specific revenue or cost synergies promised at the time of a deal, and an honest accounting of which of them actually materialized. Investors have seen enough “synergies” slides that never got tracked against reality to specifically look for the follow-up.

At pre-seed this folder might just be a handful of LOIs (letters of intent) or pilot agreements, and that’s fine. It stops being fine the moment there’s real revenue on the dashboard and nothing in this folder to back it up. That mismatch is one of the fastest ways to turn a promising Series C conversation into a skeptical one.

09_team.

Early on, an investor is backing two or three people they've gotten to know personally. By growth stage they're underwriting an organization, and they want to see its actual shape. So this folder holds the org chart, key leadership bios, employment and advisor agreements, and, as you scale, a compensation and equity summary for the leadership team. The document that matters more than founders expect is the one that maps key-person risk: what actually happens to this company if one specific person leaves. At Series B and beyond, a single un-backfilled dependency (the only person who understands the core model, or owns every major customer relationship, or holds the technical architecture in their head) is a real diligence flag, not a sign of a strong founder. The best version of this folder doesn't just list who's there. It quietly shows how the company would survive any one of them walking out the door, which is a very different and much more fundable story than "we're screwed without our CTO."

10_compliance.

This folder is comprised of three things founders usually keep in separate mental buckets: regulatory, intellectual property, and security. A due diligence process increasingly looks at these from the lens of one question: what here (in this company) is legally defensible, and what could blow up after close. On the regulatory side, investors are looking for sector-specific licenses and certifications. In the agtech and food tech space, for example, a USDA organic certification or an FDA registration isn't just menial paperwork, it's often the infrastructure supporting an actual moat. On the IP side, investors are keeping an eye out for patents and trademarks (if applicable), and clean assignment of that IP to the company. Founder-owned code or a patent that was never formally assigned to the entity is one of the most common and most fixable things that spooks a late-stage investor, and it's much cheaper to fix in month one of building the data room than in the final week of a close. On the security side, investors might ask about SOC 2 (especially for SaaS companies), a security and data-handling audit that enterprise buyers increasingly require before they'll sign a contract. Data-privacy compliance (CCPA, GDPR if you have international users) lives here too.

The reason all of this belongs in one folder is that it answers a single underwriting question: which of these is a moat, and which is a landmine. A missing certification in a regulated sector or an unassigned patent can gate a deal as hard as a bad number can.

11_governance.

This is the folder that shows you've been running a real company between board meetings, not managing by vibes. Board deck history, meeting minutes, and a clean run of prior investor updates. Individually, none of these documents is cause for drama. Collectively they tell an incoming investor something very specific: that you communicate consistently, govern deliberately, and won't go dark the moment something goes wrong.

A tidy two-year archive of quarterly investor updates is one of the quietest, most persuasive things in a growth-stage room, because it's the closest thing to a preview of how you'll treat the person about to wire you money.

This is also where references live: a short, ready list of customer and existing-investor references an incoming investor can actually call. Nobody has to label a folder "references," but the best-prepared founders have it staged the moment it's asked for. A slow, scrambling response to "can we talk to a few of your customers and your current investors?" reads as a red flag entirely on its own, regardless of what those calls would eventually have said.

Where the room moves: from Drive to a real platform

In Part I, I said a well-organized Google Drive or Dropbox folder is completely fine at the early stage, and I meant it. Here’s where that changes.

At GrubMarket, once we were deep into late-stage rounds, we moved the entire room onto Intralinks, because that’s what the Tier-1 investment banks recommended, and there’s a real reason for that recommendation. A dedicated virtual data room (VDR) platform gives you version control that actually works (no more “which of these six file names is current”), much more granular access permissions (per-user, per-folder, sometimes per-document, so a strategic investor who’s also a potential competitor doesn’t see the same commercial detail as a passive financial investor), watermarking and download restrictions, and a full audit trail of who opened what and for how long, which is genuinely useful competitive and interest-level intelligence, not just a compliance feature. When you can see that a particular fund’s partner spent forty minutes in your cohort data three times this week, that tells you something.

The honest tradeoff: platforms like Intralinks or Datasite cost real money and add process overhead. That’s exactly why they’re not needed until the scale of the round is larger, and the number of parties who need carefully controlled access justifies it. Using one at pre-seed would be like hiring a law firm to draft your roommate agreement. Not using one by the time you’re running a competitive growth or late-stage process with a real banker involved and a dozen growth equity funds in the room would be the opposite mistake. The rough line: Series B/C is when the conversation starts, especially if a banker is involved; by Series D+ it’s simply expected.

What’s still ahead

Six new folders and a platform upgrade get you a room that looks the part. They don’t, by themselves, get you through diligence. The folders in this piece are mostly qualitative: roadmaps, contracts, org charts, licenses. The part that actually decides whether a growth-stage round closes is quantitative, and it lives almost entirely inside 02_financials and the cap table.

That’s Part III: Next week, we’ll cover some grittier topics like Adjusted EBITDA and the bridge that either builds trust or destroys it, a free cash flow waterfall, net working capital, the ARR waterfall, net revenue retention and gross revenue retention broken all the way down to the actual formula, logo retention by business model, the liquidation preference stack, and the point in a live round where the term sheet itself belongs in the room. Same plain-English treatment as everything above, just with some more numbers behind it. Stay tuned.

Subscribe now

Here’s the version with the explanations attached.

A pitch deck is the version of your company you want investors to believe. A data room is the version that has to be true. Most founders spend three weeks on slide 7 of the deck and throw the rest of the data room together the weekend before diligence starts. That’s backwards, and it’s the single most fixable mistake I see in a raise. Nobody opens a cap table for entertainment. They’re checking whether the company they just heard about in the pitch is the same one sitting in the numbers. So let’s walk through the room, folder by folder, the way an investor actually would.

This is Part 1, and it’s deliberately about the core: the five folders that matter most and that any company, at any stage, needs to get right. I’ve written it with early-stage founders in mind, pre-seed through Series A, because that’s where most of you are and where the fundamentals get set. Part 2 covers everything a growth or late-stage company adds on top. But nobody should be adding the advanced folders until the core ones are clean, so start here.

01_deck: the story, with receipts attached

First thing worth noticing: it says v12. Not v1. Every founder’s deck goes through more revisions than they’d like to admit, and that’s fine. A deck that’s been stress-tested by twelve rounds of feedback is a better deck than one that’s never left PowerPoint. What’s not fine is when the version in the data room isn’t the version you actually presented. I’ve seen founders leave an older deck in the room because updating it feels like busywork. Investors notice. It reads as either sloppiness or, worse, as if the numbers changed and nobody wants to explain why.

The appendix is where the slides that would have made your deck too long go to live: the deeper competitive breakdown, the fuller market-sizing build (your TAM, SAM, and SOM, and yes, investors can tell when the top-down TAM number and the bottom-up SOM number were never reconciled with each other), the go-to-market motion broken out channel by channel, the technical architecture diagram nobody needed in the first ten minutes but somebody on the investment committee absolutely will ask about later. Sometimes this is included in the main deck as an Appendix section in the back, and that’s ok too.

The one-pager is optional, but “optional: doesn’t mean “skip it”. I highly recommend having one, but keep in mind that it’s not just a shorter markdown of the deck. It’s the thing that gets forwarded internally; to the partner who wasn’t in your pitch meeting, and to the associate deciding whether to take an intro call. Every venture firm makes decisions in a partner meeting you’re not in the room for. Your one-pager is what represents you in that room. If it’s bad, your story dies inside the firm before it reaches the people who’d actually champion it.

02_financials: where the story gets tested

This is the folder where most first-time founders underbuild, and it’s the one investors spend the most silent time in. The model needs to be stress-tested, meaning someone (you, ideally, not just a banker or a hired analyst) has actually pulled the growth assumption down and watched what happens to runway (how many months of cash you have left at your current burn) and burn rate (how much cash you’re consuming each month). Investors will run these scenarios themselves regardless. The question is whether you already know the answer when they ask “what happens if you grow half as fast as this model assumes?”

At the early stage, your model is a projection, not a track record, and everyone knows it: nobody expects three years of clean audited history from a company that’s eighteen months old. But the assumptions underneath the projection are fair game, and a model that falls apart the moment someone questions a single input tells an investor more than the headline growth number ever could.

Unit economics deserves its own file, separate from the model, because it answers a fundamentally different question: not “will this company grow,” but “does growth make the company more valuable or just bigger.” This is where the real vocabulary lives, and it’s worth being precise about it:

• CAC (Customer Acquisition Cost): the fully loaded cost (sales, marketing, the people and tools behind them) to win one new customer. Not just ad spend. The number founders quote is almost always too low because they forget to take wages, salaries and other overhead costs into consideration.

• LTV (Lifetime Value): the total gross profit you expect from a customer across their entire relationship with you. Note gross profit, not revenue: a common mistake is computing LTV off revenue and quietly inflating it by your entire cost of goods.

• LTV:CAC Ratio: how many dollars of lifetime value you get for every dollar spent acquiring a customer. Investors generally want to see a path to 3:1 or better. Below 1:1 means you’re literally paying more to acquire customers than they’re worth, which is a machine for lighting money on fire.

• Payback Period: how many months of gross margin it takes to earn back the CAC. Twelve months or under is strong for most early software businesses; the longer it stretches, the more cash you have to float before a customer becomes net-positive.

• Gross Margin: revenue minus the direct cost of delivering the product, as a percentage. A “software” company running at 40% gross margin is telling you it isn’t really a software company underneath, and investors will price it accordingly.

Cohort data is the one that’s genuinely nice-to-have at the earliest stages and increasingly non-negotiable the further along you get. A cohort is just a group of customers who started in the same period (i.e. everyone who signed up in July) tracked over time. Cohort analysis is how you tell the difference between a business that’s growing because it retains and expands what it wins, versus one that’s growing only because it’s pouring more into the top of the funnel while the bottom quietly leaks. At minimum, it’s proof that retention isn’t a story you’re telling, it’s a pattern you can show.

03_cap_table: the cleanliness test

Notice the status tag here is simply “Clean.” That’s not a throwaway label: a messy cap table is one of the fastest ways to slow down or kill a round, and it’s almost entirely self-inflicted. Convertible notes that never converted properly, SAFEs with inconsistent terms (a SAFE, Simple Agreement for Future Equity, is the standard early-stage instrument that converts to shares at your next priced round; simple in theory, messy in practice when you’ve signed a dozen with different valuation caps and nobody’s tracking what they’ll convert into), advisors who were promised equity in a Slack message and never got paperwork. None of it is fatal on its own. All of it adds weeks to diligence and, worse, makes an investor wonder what else in the room hasn’t been kept up.

What an investor is actually checking here is your fully diluted share count. Not just the shares issued today, but everything that could convert into shares: the outstanding options, the unissued option pool, every SAFE and convertible note. Founders love to quote ownership percentages off issued shares because the number looks better. Investors do the math off fully diluted because that’s what they’ll actually own after everything converts. The gap between those two numbers is where a lot of first-time founders get an unpleasant surprise about their own dilution.

The ESOP summary is where founders either show they understand their own option pool or reveal they’ve never actually looked at it. ESOP stands for Employee Stock Option Plan, the pool of equity you set aside to hire and retain your team. How much is allocated, how much is left, what the strike prices look like; an investor evaluating a round needs to know whether the pool needs refreshing (topping up) as part of the deal, and here’s the part that stings if you don’t see it coming: a pool refresh usually comes out of the existing shareholders’ ownership, meaning yours, before the new money comes in. Understanding that going in means you can negotiate it. Getting surprised by it means you negotiate it badly.

04_legal: the folder nobody wants to build and everybody needs

Articles of incorporation being marked “Signed” seems almost too basic to mention, except that I’ve watched rounds stall because a company’s foundational documents were technically incomplete: an amendment that never got filed, a state registration that lapsed, a Delaware franchise tax nobody paid. It’s boring. It’s also table stakes, and the fact that it’s boring is exactly why it gets neglected.

IP assignments matter more than founders expect, especially for anything technical. An IP assignment is the document that transfers ownership of intellectual property (code, designs, inventions) from the individual who created it to the company. If a co-founder, an early contractor, or a friend who “helped out for a few weeks” wrote code or built something that ended up in the product without a signed IP assignment, then legally, the company may not actually own its own product. That’s a real gap an investor’s counsel will find, and it’s a much cheaper problem to fix at incorporation than to untangle three years and two rounds later when the person who wrote your core module is unreachable or unhappy.

Early on, articles and IP assignments are close to the whole folder, and that’s appropriate. This is one of the folders that grows the most as you scale (Part 2 covers what gets added), but the foundation being genuinely clean matters at every stage, and it’s easiest to get right when the company is small.

05_metrics: the one that has to stay honest

“Live” is doing real work as a status tag here, not just as a design flourish. A metrics dashboard that was accurate three months ago and hasn’t been touched since is worse than no dashboard at all, because it invites the obvious question: why hasn’t this been updated, and what changed that you’d rather not show? The fix isn’t complicated. It’s just discipline.

At the early stage, your KPI dashboard should show whatever your real north-star metrics actually are, presented honestly: active users, activation rate (the percentage of signups who reach the moment where they actually get value from the product), month-over-month growth, early retention. You don’t need the full late-stage metrics vocabulary yet (that’s Part 2), but you do need the numbers you do have to be real, current, and consistent with what’s in your model and your deck. The single fastest way to lose an investor’s trust is a metric that says one thing on the dashboard and another thing in the financial model.

The sales pipeline is your list of prospective deals and where each one sits in the process. At pre-seed, growth is a hypothesis, and the pipeline is mostly a signal of demand (i.e. are people actually trying to buy this?). By Series A it starts becoming evidence of a repeatable motion.

Where the room lives at the early stage

One thing nobody tells first-time founders: the platform your data room lives on is its own decision, and at the early stage it’s a much simpler one than people make it out to be.

At pre-seed through Series A, a well-organized Google Drive or Dropbox folder is completely fine. I’ve seen plenty of clean, fast, successful early-stage raises run entirely out of Drive. Nobody is judging you for not paying for enterprise software when you’re an eight-person company. What actually matters at that stage is the organization inside it: consistent file naming, a clear folder structure that mirrors what we just walked through, and permissions that don’t accidentally give a prospective investor edit access to your cap table (set everything to view-only, and share folders, not your entire Drive).

There’s a whole world of dedicated virtual data room platforms, like Intralinks and Datasite, with version control, granular per-user access, watermarking, and audit trails. They’re genuinely valuable, and they’re also genuinely unnecessary at the early stage. Using one at pre-seed would be like hiring a law firm to draft your roommate agreement. I’ll cover when that shift actually makes sense in Part 2.

The thread that ties it together

Every folder in this core room is answering the same underlying question, just from a different angle: is the company in front of me the same company on paper? The deck tells investors who you say you are. The financials show whether the growth is real and whether it makes the company more valuable or just bigger. The cap table shows whether you’ve run a tight operation or a messy one. The legal folder shows whether the foundation is actually solid or just looks that way from the pitch. The metrics show whether you’re willing to let them see the truth in real time, not just the truth as of your last board deck.

Get these five right and you’ve built a room that can carry an early-stage raise. Build the room like someone smarter than you is already inside it, because by the time it matters, they are. A pitch deck earns you a meeting. A data room earns you the wire.

Next week, in Part 2, I’ll walk through what a data room grows into: the additional folders (product, commercial, regulatory, and more) and the heavier financial machinery (Adjusted EBITDA, free cash flow waterfalls, net revenue retention, the rule of 40) that growth and late-stage investors expect once you’re asking them to underwrite an outcome instead of a story. The core stays the same. The room around it gets a lot bigger.

I’ve spent my career on the finance and storytelling side of startups. Investor relations, fundraising, VC and founder coaching. I have never written a line of production code, and six days before the site went live I couldn’t have told you what a git commit was, why anyone would host on Vercel, or what a CLAUDE.md file did.

The site you’re looking at, I built and shipped myself. No developer, no agency, no template I bought and filled in. It took 35 hours spread across 6 days, and by the end I understood the whole pipeline well enough to migrate my own domain between hosts without help. Here’s exactly how, prompt by prompt, so you can try it yourself.

I’m writing this down for two reasons. One, to help others (and the founders I work with) realize that the barrier to entry to building something “premium” is lower than ever. Two, it’s a decent first look at how I actually work: I care more about the story a thing tells than about doing it the hard way. This is the real teardown, not a highlight reel. Every major prompt, why I made the design calls I made, where I got stuck, and the whole receipt for my website.

Where the 35 hours actually went

The build wasn’t 35 hours of typing prompts. Most of it was learning. Roughly how it broke down:

• Getting fluent in Claude Code. The first several hours were just me learning how to talk to it, what a terminal is, how sessions and context work, and how to read what it was doing instead of blindly accepting its edits.

• Finding the animations. A surprising amount of time went into hunting for motion I actually liked, on sites like 21st.dev and GSAP demos, and figuring out what was realistic to pull off.

• Fighting context windows and usage limits. Managing Claude Code’s memory across a multi-day build is a real skill, and it cost me time until I got the hang of it. More on this below.

• Iterations and bug fixes. The long tail. Spacing that broke, an animation that fired before the page loaded, mobile layouts that looked nothing like the desktop version.

• Feedback rounds. I sent early versions to friends, got honest notes on flow and content, and reworked whole sections based on what they said.

The stack, in plain English

Four tools did all the work. None of them require you to already know how to code, but I’ll give you the real terms alongside the plain-English version, because knowing the vocabulary is half of feeling in control when learning something new.

Claude Code is the builder. It’s an AI you talk to in your terminal in plain English, and it writes the actual files, installs what it needs, runs commands, and fixes its own mistakes when you point them out. This is the program doing the heavy lifting.

Cloudflare is where I bought the domain. It’s a domain registrar, so jonathanhua.com started life as a purchase in my Cloudflare account.

GitHub is the save button, or in real terms, version control. Every change gets committed to a repository with a full history, so nothing is ever lost and you can roll back to any previous state. Think of it as version history for the whole project instead of one document. It’s also the handoff point to the host.

Vercel is the publish button, or the host. It takes the code from GitHub and puts it on the internet by pointing the code to a real URL. I chose Vercel specifically because the site is built in Next.js, and Vercel is the company that makes Next.js. That means zero-config deploys and first-class support instead of fighting my tools.

How the pipeline comes together: I change code in Claude Code, push it to GitHub, and Vercel automatically builds and deploys it. I’ll break that whole chain down at the end, because understanding it is the difference between “I made a website” and “I run a website.”

Step 1: Getting Claude Code running

This was the only part that felt technical, and it still took about ten minutes. I installed Claude Code, opened the terminal, typed claude, and started talking to it like a person.

My honest first message was close to this:

I’ve never built a website and I don’t write code. I want to build a personal site for my advisory business at jonathanhua.com using Next.js. Before you do anything, explain what you’re about to do in plain English, and tell me what I need installed. Don’t assume I know developer terms.

That instruction, “explain it to me like I don’t know developer terms,” did more work than anything else I typed all week. I reused it constantly.

Step 2: Giving it a brain (the CLAUDE.md file)

If you take one technical concept from this post, take this one.

CLAUDE.md is a plain markdown file that lives at the root of your project. Claude Code reads it automatically at the start of every single session, before it does anything else. Here’s why that matters: each Claude Code session starts with a fresh context window, meaning it forgets the specifics of your last conversation. CLAUDE.md is how you stop re-explaining your project every time you sit down. It’s persistent memory.

I generated a first version by running /init, which analyzes the project and writes a starter file for you. Then I rewrote it by hand until it actually described my project and my taste. Mine ended up looking roughly like this:

# jonathanhua.com

## Project
Personal site for Jonathan Hua — fundraising, IR, storytelling, founder coaching.
Multi-page (Home, About, Publications, Podcast, Work With Me).
Audience: founders and investors. Job: establish credibility and drive contact.

## Tech stack
- Next.js 14 (App Router) + TypeScript + Tailwind CSS
- Framer Motion for animation (scroll reveals, hero name reveal, count-up stats)
- react-simple-maps for the SVG journey map on the About page
- Contact form: Web3Forms (no backend — posts straight to my inbox)
- Deployed on Vercel, source on GitHub

## Design system  (colors stored as RGB channels so Tailwind opacity works)
- Background: warm ivory   #F6F2E9
- Text: near-black ink      #1E1A15   / muted #5E5649
- Accent (CTAs, links):     copper    #AD5929
- Secondary accent:         olive     #4D5B39
- Borders / dividers:       line      #DFD7C6
- Type: DM Serif Display for headlines, Manrope for body
- Voice: confident, senior, understated. Numbers over adjectives.
- Lots of whitespace. Nothing flashy or bouncy.

## Rules (read these first)
- public/ is ALWAYS lowercase. Capital “Public/” works on my Mac but breaks
  every image on Vercel, because Linux is case-sensitive. This cost me a whole day.
- Default to LIGHT mode. Dark mode is opt-in via the toggle, class-based,
  persisted in localStorage. Never auto-apply dark.
- All content is hardcoded in components/sections/. There is no CMS.
- I write my own bio and section copy. Do NOT write my copy.
- Mobile and desktop are optimized separately, not just scaled.
- Keep animations subtle. Respect prefers-reduced-motion.
- Explain technical steps in plain English before running them.

Over the build this grew into a much longer “project bible” — the full file structure, every component, and a running log of every bug and its fix — but the block above is the core of it, and the core is what does the work.

A few things I learned about writing a good CLAUDE.md: keep it high-signal, because a bloated file eats context and the model follows it less reliably. Write rules as direct commands, not suggestions. Include a “what to avoid” section, because telling it what not to do matters as much as what to do. And it survives compaction, so even when a long session gets compressed, Claude Code re-reads CLAUDE.md from disk and my instructions stay live.

Step 3: Setting the design direction

I didn’t design anything up front, and I also didn’t just let Claude reach for its defaults. Left alone, AI builds sites that all look the same. There’s a recognizable “an AI made this” aesthetic, and I did not want my professional front door wearing it.

So before building, I used Claude Code’s frontend-design skill to force deliberate choices. A skill is a set of expert instructions Claude Code loads for a specific kind of task. This one pushes it to act like a design lead who makes intentional, opinionated calls on palette, typography, and layout instead of grabbing a template. My prompt was roughly:

Use the frontend-design skill. Treat this like a real brief. Before you build anything, give me a design plan: a named color palette in hex, a display and body typeface pairing, a layout concept, and one signature element the site will be remembered by. My positioning is “both sides of the table, every stage of the raise.” It should feel senior and confident, not startup-flashy.

That produced an actual design system I could react to, not a random first draft. The choices I landed on, and why:

• Minimalism and heavy whitespace. My whole pitch is credibility and judgment. A calm, uncrowded page reads as senior. Busy reads as junior.

• A warm, restrained palette. An ivory background (#F6F2E9), a single copper accent (#AD5929), near-black ink for text (#1E1A15). One accent color used sparingly reads more confident than a rainbow.

• DM Serif Display over Manrope. An elegant serif for the headlines and the styled moments, a clean geometric sans for everything else. The pairing does a lot of the “senior” work on its own.

• A four-part service grid: Fundraising, Investor Relations, Storytelling, Founder Coaching. Clear, scannable, and it maps exactly to how I actually work.

• A stats banner with the hard numbers: close to $300M raised, a 3.5x average valuation uplift, 10+ years in VC and tech. Numbers carry credibility better than adjectives ever will, so I let them do the talking — and they count up as the band scrolls into view.

• A quiet, confident hero: my name, then one line, “Raising capital is a skill, and I teach it.” No stock photo. No shouting.

Step 4: Building it, section by section

From here it was a conversation. I’d describe a section, look at what it made, and correct it. The pattern that worked: be specific about the feeling and the priority, and let Claude handle the how.

Scaffolding the project.

Scaffold the Next.js project with Tailwind. Set up the folder structure and a single scrolling page with empty sections in this order: hero, about, services, stats, contact. Get it running locally so I can see it in my browser.

Within a few minutes I had a rough site running on my own machine that only I could see. Ugly, but real.

The hero.

Build the hero. My name large at the top, then the line “Raising capital is a skill, and I teach it.” underneath, smaller. Enormous amounts of empty space. It should feel like someone who doesn’t need to fill the screen to be taken seriously.

The services grid.

Build the services section as a four-item grid: Fundraising, Investor Relations, Storytelling, Founder Coaching. Each gets a short one-line description underneath. Keep the type consistent with our design system. On desktop it’s a row of four; we’ll handle mobile separately.

The stats banner.

Add a stats band between about and services. Three or four numbers with tiny labels: “~$300M raised,” “3.5x valuation increase,” and the others I’ll give you. Big numbers, small labels, no icons, no clutter. Let the figures carry it.

The copy stays mine. For every text-heavy section I told it explicitly: build the layout, leave the words to me. The tools can build the room. The story on the walls is the whole reason anyone hires me, so I wasn’t about to outsource it.

By day two or three my prompts had noticeably sharpened. Early on I’d say “make it look nicer” and get nothing useful. By the end I was saying “reduce the vertical padding on the services section by about a third and left-align the descriptions,” and getting exactly that. Being specific about the feeling and the priority, then letting Claude handle the how, beats “make it look professional” every time. Learning to give precise, constrained instructions was most of the skill.

Step 5: The part that made it look expensive (motion)

A static site looks fine. A site where things move as you scroll looks like someone spent money on it. Most of that motion on my site — sections fading and rising into place as you scroll, my name revealing on load, the stats counting up — runs on Framer Motion, an animation library made for React, which is what Next.js is built on. It’s the reason the whole thing feels alive without feeling busy.

I also layered in a little bit of GSAP (the GreenSock Animation Platform, the engine behind a huge share of award-winning sites) for one or two micro-interactions, like the scroll-progress line along the top edge. Worth knowing if you go deeper on GSAP: it used to be a paid product for the good stuff, but Webflow acquired GreenSock and made the entire library, including all the premium plugins, 100% free in April 2025, commercial use included. The polished effects you used to only see on agency sites are now free to anyone.

I didn’t have to learn either library to make use of them. Instead, I just described the motion I wanted:

Add subtle scroll animation. As someone scrolls, each section should fade and rise gently into place rather than just appearing. On the hero, animate my name in word by word on page load. Install what you need. Keep it tasteful. Nothing bouncy. And respect prefers-reduced-motion.

Claude wired up the scroll reveals and the hero animation and wrote all the code. My job was to react: “too fast,” “too much movement,” “delay the hero by half a second.” We went back and forth on timing more than any other part of the build, maybe a dozen passes, because subtle is much harder to get right than dramatic. This is the step that took the site from “a guy made this” to “wait, who built this for you.”

Step 6: Not building components from scratch (21st.dev)

The move that saved me the most time: instead of asking Claude to invent every button, card, and layout block from nothing, I pulled polished ones from 21st.dev.

21st.dev is a big community registry of ready-made React components, built on the same Tailwind stack my site uses. You browse it like a catalog, find a component you like, and it comes as real code you own and can change, not some locked-down package. It also has an AI feature where you describe a component and it generates a few versions to pick from. Because it’s the same stack, the components drop cleanly into a Next.js project.

My workflow was simple. I’d find something on 21st.dev that had the look I wanted, then hand it to Claude:

Here’s a component from 21st.dev I want for my services section. Add it to the project, restyle it to match our existing palette and typefaces, and swap in my real content. Don’t let it break the design system we set up.

Starting from a designer’s component and adapting it beats generating from scratch every time, and it keeps the whole site consistent instead of looking like five different people built it.

Step 7: Desktop and mobile, optimized differently

Most people scale a desktop site down and call it “mobile.” I didn’t want that. I told Claude Code to treat them as two different experiences:

Now optimize the mobile experience separately. Don’t just shrink the desktop layout. Collapse the four-part services grid into a horizontal scrollable single layered column, reduce the type scale, tighten the section spacing, and simplify the animations so they’re smooth on a phone. Make sure tap targets are big enough and nothing overflows horizontally.

This is where responsive design and breakpoints came in. Tailwind uses breakpoints (the screen widths where the layout changes), so the same component can render as a four-across row on a laptop and a clean vertical stack on a phone. I checked every section at mobile width and sent it back whenever something looked cramped or an animation stuttered on a smaller screen.

Step 8: Putting it on the internet

This is the part that made me feel like I actually run a website instead of just having made one. Here’s the whole chain.

1. The domain lived on Cloudflare. I’d bought jonathanhua.com through Cloudflare’s registrar, so that’s where DNS started.

2. The code went to GitHub. I had Claude Code initialize a git repository, then create a repo on GitHub and push to it. In practice I just said “set up git and push this to a new GitHub repo,” and it ran the commands: git init, staged the files, git commit, connected the remote, and git push. GitHub is now the source of truth for the entire site. Every change I make is a new commit, so I have a complete, reversible history. Connecting Claude Code to GitHub meant my local work and the cloud repo stayed in sync with a single instruction.

3. Vercel hosts it, straight off GitHub. I connected the GitHub repo to Vercel. Vercel auto-detected that it’s a Next.js project and configured the build with zero setup, which is exactly why I chose it for a Next.js site. Now the flow runs itself: I change something in Claude Code, push the commit to GitHub, and Vercel sees the new commit, rebuilds the site, and deploys it live in about a minute. That’s a real CI/CD pipeline (continuous integration and deployment), and I have one running for a site I built without knowing how to code a week earlier. Push to the main branch and it goes to production; every branch gets its own preview URL so I can see changes before they’re live.

4. Then I moved the domain to Vercel. Most recently I transferred the domain itself from Cloudflare into Vercel, so my registrar, DNS, and hosting all live in one place. That meant unlocking the domain at Cloudflare, grabbing the transfer authorization code, initiating the transfer on Vercel’s side, and pointing the nameservers. One less dashboard, one less thing to keep in sync.

So now, every time I want to change something, I tell Claude, it updates the files, I say “push it live,” and about a minute later the real site at jonathanhua.com has the change. No agency, no waiting, no invoice.

The unglamorous skill: context windows and usage limits

Here’s the part nobody puts in the “I built a site with AI” posts.

Claude Code has a context window, a finite amount it can hold in working memory at once. On a build this size, spread over days, you hit its edges. Long sessions fill up and get compacted, and if you’re not deliberate, the model loses the thread. Here’s how I learned to manage it:

• I leaned on CLAUDE.md as durable memory, so even after a context reset, my project rules reloaded automatically and I didn’t have to re-brief it.

• I used /clear to wipe the slate between unrelated tasks in the same session, which reloads CLAUDE.md fresh and stops old context from muddying a new problem.

• I kept prompts focused and referenced specific files instead of making it re-scan everything.

• I ran into usage limits and learned to pace the work across the six days, tackling bigger builds when I had usage space and saving small tweaks for when I didn’t.

Learning to work with the context window instead of against it was one of the most genuinely useful things I picked up, and it’s the kind of thing that only shows up when you build something real and substantial.

Getting honest feedback

Once it looked right to me, I sent it to a select handful of friends and asked for real notes on flow and content, not just compliments. And boy, did they give me pointers…

“A couple of sections were in the wrong order. One line I loved was confusing to everyone but me. The flow from hero into services had a dead spot…”

I took the notes back to Claude Code, reordered sections, rewrote copy, and tightened the transitions. The site got meaningfully better because I asked people to be honest and then actually acted on it. Same advice I give founders about their pitch: your favorite line is often the one to cut.

What actually broke

It wasn’t all clean. The worst one looked like a catastrophe and turned out to be almost stupidly simple. Every image on the site — my portrait, the portfolio logos, the press logos — loaded perfectly on my laptop, and then vanished the second I deployed to Vercel. Broken-image icons everywhere on the live site, while localhost looked flawless. The cause: I’d named my image folder Public/ with a capital P. macOS doesn’t care about capitalization, so it worked on my machine; Vercel runs on Linux, which very much does care, and Next.js expects a lowercase public/. Renaming one folder fixed every image at once. I would never have guessed that on my own, and it’s now the first rule in my CLAUDE.md.

A few others, all in the same spirit. The site kept opening in dark mode even though I’d designed it light-first — some leftover logic was auto-applying a dark theme on load, and I had to have Claude rip it out so light became the true default and dark stayed opt-in. On the About page, the animated copper underline under my bio kept breaking into a row of little dashes as it drew itself in, instead of inking as one clean continuous line; that one took a few passes to actually fix. The publications heading wrapped onto two lines on mobile and threw off the whole layout. And the contact form was its own small saga: I started with no form, added Formspree, bumped into its free-tier limits, and finally moved to Web3Forms, which has much higher submission limits on its free-tier and posts straight to my inbox with no backend at all.

The fix was almost always the same loop: paste the error or describe what looked wrong, and let Claude debug its own work. I never once opened a Stack Overflow tab. Debugging turned out to be a conversation, not a research project.

Why I’m telling you this

I don’t think the lesson here is “everyone should build their own website.” It’s narrower and more useful than that.

The thing standing between an idea and a real, working version of it used to be a skill you either had or paid a lot for. That wall is mostly gone, and getting over it is now mostly about knowing how to direct these tools well: writing a tight CLAUDE.md, giving precise constrained instructions, managing context, and understanding your pipeline enough to fix it when it breaks. Those are learnable in days, not years. I’m proof.

So when a founder tells me they can’t test their idea until they raise enough to hire an engineer, I politely tell them their limitation is their own imagination. Start. Describe what you want in plain words. Correct it when it’s wrong. Ship the ugly version and improve it in public. That’s the same advice I give founders about their fundraise story, and it turns out it’s the same advice for building the thing that tells it.

The site is at https://www.jonathanhua.com. This build was step one. I’ve got more AI projects underway, and I’ll be writing them up here as I go. If you’re a founder trying to build or raise and you want a second set of eyes, that’s what I do. Come say hi.

Subscribe now

If this was useful, subscribe. I write about fundraising, IR, storytelling, building with AI, and the future of food, roughly whenever I have something worth saying.

Prior to Bessemer, Aditya worked in consulting at McKinsey & Company as well as in business strategy and sales operations at SpotHero, a parking technology and marketplace company.

Aditya graduated from Northwestern University with a degree in economics and two minors in business and entrepreneurship. He also is the founder of Rock Venture Capital, Northwestern’s undergraduate level venture capital organization. Outside of work, Aditya enjoys playing soccer and cricket, is a diehard Formula 1 fan, and has a keen interest in aviation. He plans on getting a pilot’s license in the future.

Companies are increasingly adapting to a pandemic that has dramatically altered how employees interact, train, and learn new skills. This change is particularly pronounced in the healthcare space where surgeons currently have limited training options if expensive equipment like cadavers or simulators are required, and they end up relying on written instructions. This has made current medical training less effective and is exasperated because hospital budgets are more strained than ever due to Covid and have less funding for training opportunities.

However, training surgeons remains essential to ensure patients have the best outcome possible. As of now, 400K patients each year experience preventable harm, and the mortality rates of patients with lower-skilled surgeons is 5x higher than patients with highly skilled ones. In addition, without reinforcement training, nearly 31% of new surgeons have trouble properly completing surgical procedures and need supplemental experience. A new cost-effective solution to surgeon training is using VR. By creating a virtual operating room, surgeons can learn and train procedures in a risk-free environment which allows for ample training before meeting a live patient. Having VR training during Covid is important too as in-person courses are limited.

This is why we recognized the importance and potential of Osso VR and are excited to support them in their Series A. In addition to quality virtual training, the Osso VR platform helps hospital staff assess surgeon skills, and surgical teams more efficiently and effectively utilize new surgical technologies. Osso VR’s platform intelligently combines the Oculus Quest with a proprietary VR content creation engine that merges a content library with analytics dashboards to track the performance of each user. The long-term mission is to enable a wide range of medical professionals to succeed through remote training and to help increase adoption of higher-value medical technologies. We have been tracking mixed reality technologies for several years now, but only recently became confident about the market timing of VR after we identified healthcare and surgical training as emerging use cases for this kind of technology.

All-Star Founder

When I met founder Justin Barad for the first time, I was immediately impressed. Like many investors, we look for founders who display a certain kind of grit, deep domain expertise, a proactive (rather than reactive) leadership style, and a propensity to succeed under pressure. Justin has a solid combination of credibility in healthcare and an understanding of tech and gaming. Before starting Osso VR, Justin was a game developer and owns a game development credit from Activision/Blizzard. However, when a family member was hospitalized, Justin realized that hospitals were drowning under antiquated technologies and ineffective software solutions that were creating barriers for doctors to provide the best care to patients. This drove him to attend medical school and ultimately build Osso VR.

What is remarkable about Justin’s entire career trajectory is his ability to excel at everything. Armed with a bioengineering degree from UC Berkeley, Justin got an MD from UCLA’s Geffen School of Medicine where he graduated first in his class and stayed there for his orthopedic surgery residency. He also did fellowships at Harvard and Boston Children’s Hospital, and spent time as a Fellow at Stanford Byers Center for Biodesign. He is now a highly sought-after pediatric orthopedic surgeon and has built an impressive team around him at Osso VR.

Justin has demonstrated a track record of success across multiple types of roles and jobs which gives us confidence in his ability to learn quickly, iterate on the product, drive Osso’s go-to-market strategy and succeed in building a solution with strong product-market fit. But we’re not the only ones impressed. In the middle of fundraising, Justin managed to add both ILM alum Art Director Jonathan Sabella (who happened to supervise the first team to win an Oscar for a VR film) and seasoned healthcare executive Josh Willeford (CFO) to his team.

Strong Path to Market, Incredible Results

The first time we visited Osso VR, we had the opportunity to perform an intramedullary tibial nail insertion surgery in Osso’s virtual operating room. The setup and calibration were quick, and about a minute after I put on the headset, I was already assembling drivers, hammering a tibial nail into a virtual knee and learning about new tools like proximal locking jigs and stardrive screwdrivers. To this day, I can vividly remember how I felt being guided through each step of the procedure and how the real-time feedback was helping me. Though I only got a grade of ‘B’ on my very first attempt, I instantly understood why Osso’s clients, as well as surgeons who have tried the experience, value the platform.

They currently have traction with over 30 paying customers, including 12 of the largest medical device companies worldwide. Medical device training is a market that is worth over $4B globally and major medical device companies spend nearly $200M/year each on physician training. Choosing to distribute their product through medical device company sales reps creates a faster path to market, enabling Osso VR to reach hospitals and clinics more efficiently. With this go-to-market strategy, they already have captured a majority of the remote surgical market share. On top of that, Osso VR is also deployed in teaching hospitals in over 20 countries around the world with proven ROI to support quick growth. In fact, clinician tests show 3x improvement in surgeon efficacy and 7x improvement in new sales productivity for medical reps after using Osso VR. In addition, research published in the Harvard Business Review shows eye-popping results like 230% performance improvements, 20% increase in procedure speeds, and 38% increase in correctly completed steps for surgeons who trained using Osso VR compared to those using traditional training methods.

Final Thoughts

The usefulness and effectiveness of Osso VR’s platform have delivered significant customer growth for them. They provide strong value for medical device companies where they can significantly reduce the cost of training surgeons and also help shorten the sales cycle with healthcare providers which is notoriously tough and long-winded. Since they completed fundraising, Osso VR has continued to build on their success — they expanded their content creation team and work beyond orthopedic surgeries into other divisions like endovascular, thoracic, and even OB/GYN surgeries. They recently also upgraded the visual UI and UX of the immersive experience making it even more realistic. The data Osso VR is collecting from each training session will enable them to better customize future training experiences and build the most robust database of medical training performance and progress in the market.

The current remote training market is growing yet fragmented in healthcare and I believe that, with their functionality and innovative approach to medical training, Osso VR has the opportunity to be transformational. They are tackling a universal problem in the healthcare training space and are building a VR platform that makes healthcare training scalable. With Osso VR, millions of doctors will be better prepared and more effective, and even during Covid, can continue training.

Learn more about Osso VR or schedule a demo here.

Jonathan is a VC investor dedicated to helping impact-driven entrepreneurs scale and grow their business. Jonathan is regularly invited to host pitch competitions, at events like the Forbes Summit and SXSW, and speak on-stage about his experiences in VC, entrepreneurship, the Silicon Valley startup landscape, and his interest in the AgriFood space. In addition, Jonathan has been invited as a speaker at the Forbes Under 30 Summit, was selected as part of the 2019 AgGrad 30 Under 30 class, is an advisory council member for Gold House’s Gold Rush Accelerator, and hosts a podcast called Capitalist (Ad)Ventures that explores the diversity of thoughts and backgrounds in the entrepreneurship space. Jonathan has a BA in History from Rice University and an MBA from the Johnson School at Cornell University.

Connect With Me: LinkedIn| Medium | Email

Prior to SocialFlow, Kleida was an analyst at Digitas, where she oversaw the utilization of new technology tools to enhance performance, including advanced analytics and the creation and implementation of digital content across social channels.

Kleida graduated cum laude from Mount Holyoke College with a BA in Economics and Mathematics and holds an MBA with honors from the Kellogg School of Management.

Introduction

We are experiencing an unprecedented time in history with the threat and upheaval caused by COVID-19. While there is much fear, there are also many heroic stories and incredible efforts by strong and compassionate leaders to help us all get through this.

We are proud and heartened to see companies in our portfolio rise to the occasion making swift changes so they can contribute to COVID-19 relief efforts by providing services to support healthcare providers, patients, essential businesses, and critical supply chains. We want to give a shoutout to some of the great work our portfolio companies are doing during this pandemic.

Portfolio Company COVID-19 Initiatives

ORIGIN

San Francisco, California / 3D Printing, Manufacturing

What they are doing:

Origin worked with Harvard’s Beth Israel Deaconess Medical Center to run clinical trials and validation for creating COVID-19 test swabs with their large-scale 3D printing resources. Normally geared towards automotive and footwear products, they now dedicate 100% of their production to the test swabs that are complex to make because they require a rigid handle and flexible neck. They are partnering with Stratasys to print and distribute millions of swabs.

Considering widespread testing for COVID-19 is one of the most effective ways to manage and control the outbreak, Origin’s work is key to the healthcare industry’s ability to respond quickly and effectively to the current crisis. If you need test swabs, check their site here.

SPIRE HEALTH

San Francisco, California / Wearables, Healthcare

What they are doing:

Spire Health’s tags are regularly used to monitor sleep patterns, stress indicators, and pulse rates. People wear them to track themselves or patients use them to send data to their doctors for remote monitoring. With COVID-19, these tags are being offered to large outpatient pulmonary groups and critical care specialists to help monitor their patients while keeping them safe at home, thus reducing risk for infection. In addition, Spire Health is partnering with hospitals to use the tags to detect elevated respiratory rates in patients who may be experiencing early signs of symptoms which are often underreported. This helps with early intervention for at-risk patients.

You can read more about Spire Health’s initiative here.

Miles

Redwood City, CA / Transportation, Lifestyle, and Rewards

What they are doing:

Miles is a platform that rewards users like airline miles but for all modes of transportation. It rewards more for green modes, such as walking, and rewards all the time whether commuting, shopping, or traveling. Rewarded miles can be redeemed for a variety of products. However, COVID-19 is forcing us to stay home which makes it difficult to earn any miles.

The Miles team got creative and first launched a charity campaign where the Miles community could donate their miles for Feeding America with a target of 5,000 meals. In the end, they were able to donate 10,000 meals. With that success, they created a new campaign that successfully donated $10,000 worth of supplies, such as N95 masks and PPE, to the Good360 charity organization. And now, they have changed how they approach and incentivize users by rewarding x20 miles for walking, running, or biking if they also maintain social distancing. Read the details here.

Zippin

San Francisco, California / Retail, Automation

What they are doing:

Zippin provides brick and mortar retailers with cashierless checkout technology. Using white label modular systems, the Zippin technology is reasonably priced and can be installed in 2 weeks, making the cashierless experience accessible to small to medium businesses.

Food stores and their employees are on the front line as essential workers in a risky environment. Moving quickly to cashierless checkout is positive for everyone as it decreases the number of staff required and overall person-to-person interaction. To encourage this shift, Zippin is waiving setup fees for any retailer who wants to use their technology for stores selling essentials and safety products. If cashierless checkouts felt odd to consumers before, this experience will help shoppers understand the value. Read more about the initiative here.

Littlstar

NYC, NY / Entertainment, VR/AR, Consumer Media

What they are doing:

Littlstar provides on-demand premium consumer content in immersive virtual reality and augmented reality platforms. During this crisis, they have donated 50 headsets to senior healthcare facilities for therapy, targeting seniors who are suffering from Alzheimer’s and dementia. With families unable to visit such facilities, Littlstar is helping combat isolation.

They also have two new content initiatives in response to COVID-19. First is they are live streaming concerts to TVs, Mobile, Web, and VR with Digital Mirage raising $300,000 for musicians who need help during this pandemic. They are planning more live streaming and concerts on the platform. Second, they have introduced fitness and wellness content to help people stay healthy and energized within the confines of their homes.

As more people are turning to VR/AR to keep themselves entertained, Littlstar plans to provide meaningful entertainment through its streaming platform.

LemonBox

Beijing, China / Health, Wellness, Nutrition

What they are doing:

LemonBox is an e-commerce startup in China that creates personalized vitamin and supplement packs for consumers and provides targeted nutritional advice. As China was the epicenter of the global outbreak, LemonBox acted quickly to support local healthcare operations. In February, LemonBox took the initiative to donate 2000 N95 masks to hospitals in Hunan, China. They also started a promotional campaign by including an immune system vitamin pack for orders over $15 and have since sent out 1000 orders of immune system booster combo to people in Hubei province which is one of the areas that was hit hardest by COVID-19.

With no vaccines for COVID-19 currently available, individuals often have to take precautions and manage their immune health to decrease the risk of developing serious reactions to the virus. If you are interested in their vitamins, you can order from here.

Empowerly

San Francisco, CA / Education, Information Technology, Data Analytics

What they are doing:

Empowerly is an online student counseling platform that has built a strong community through their innovative and personalized approach. They offer guidance to students entering high school all the way to graduates looking for their first job, focusing on each student’s strengths and weaknesses beyond just academics.

Since COVID-19 hit just when students were deciding on which schools to go to or finishing up their academic year, the Empowerly team knew they had to find a way to make counseling more accessible. In order to achieve this, Empowerly is currently offering students a free digital toolkit with complementary services, which includes full access to all databases and tools, a free essay edit service, a free school research request, one-on-one discussions with former UC Berkeley admissions staff members and access to focus groups and communities on the platform. Sign up for the digital toolkit here and use the code “MEDIUM2” to get free access to two group sessions.

Coast

San Francisco, CA / Mobile, Collaboration, Remote Teams

What they are doing:

Coast is a messaging and collaboration platform for the frontline remote workforce. As more businesses consider implementing permanent flexible work schedules due to COVID-19, team communication becomes more critical than ever. This is especially applicable to healthcare teams on the frontlines that are scrambling to maintain consistent communication channels and operational workflows due to the overwhelming number of COVID-19 cases they are treating. In order to help teams like these, Coast is providing free access to their full service and platform offerings to teams in need.

In addition, Coast wants to help save local restaurants and bars that have been hit hard by quarantine orders. To do this, they built the Takeout COVID initiative, which is a public service tool aiming to connect local businesses to their communities. They have active lists for 26 cities and counting, so if you’re looking for food, wine, and drink options for delivery and takeout, check their list and support a small business.

Noom

NYC, NY / Healthcare, Wellness, Nutrition

What they are doing:

Creating long-term habits for a healthy lifestyle is challenging. Noom is an award-winning behavior change program that encourages healthy eating and exercise habits through personalized, remote, one-on-one guidance from health coaches. In our current world, where the “new normal” involves social distancing and less active lifestyles, it will be more important than ever to develop sustainable habits for overall health and wellness. In this crisis, Noom remains committed to helping people by currently offering the Noom service free of charge for two full weeks. To take advantage of their at-home curriculum and virtual services, you can learn more here.

Conclusion

We work with incredible founders in our portfolio and continue to be moved by their resourcefulness, strong leadership, and commitment to helping their communities. We remain committed to investing in and supporting ambitious and globally-focused founders with a vision to build a better future. If you are interested in learning more about or want to support these companies, please let us know at contact@scrum.vc

Jonathan is a VC investor dedicated to helping impact-driven entrepreneurs scale and grow their business. Jonathan is regularly invited to host pitch competitions, at events like the Forbes Summit and SXSW, and speak on-stage about his experiences in VC, entrepreneurship, the Silicon Valley startup landscape, and his interest in the AgriFood space. In addition, Jonathan has been invited as a speaker at the Forbes Under 30 Summit, was selected as part of the 2019 AgGrad 30 Under 30 class, is an advisory council member for Gold House’s Gold Rush Accelerator, and hosts a podcast called Capitalist (Ad)Ventures that explores the diversity of thoughts and backgrounds in the entrepreneurship space. Jonathan has a BA in History from Rice University and an MBA from the Johnson School at Cornell University.

Connect With Me: LinkedIn| Medium | Email

Prior to joining Learn, Susan spent three years at the Strategy & Ventures group of Sesame Workshop, the media and education non-profit creator of Sesame Street. There, she worked on growth and strategic planning projects, and helped manage Sesame Ventures, the VC arm of the company that invests in mission-aligned early stage kids-oriented startups.

Susan began her career in investment banking at Credit Suisse after earning her BCOM and BA in Math at Queen’s University in Kingston, Canada. She has since worked in Toronto, Shanghai, New York City, and San Francisco.

In her spare time, Susan enjoys reading, learning, running, singing, and good journalism.

Emily graduated summa cum laude from the University of Pennsylvania’s Vagelos Program in Life Sciences & Management, where she earned a B.S. in Finance from Wharton, a B.A. in Biology, and a certificate in healthcare policy as a Public Policy Research Scholar. During her time at Penn, Emily engaged in research on Castleman Disease with Dr. David Fajgenbaum, served as a project leader for 180 Degrees Consulting, and conducted due diligence on healthcare startups as part of the Wharton Impact Investing Partners health & wellness team.

Kookai is an active speaker, mentor, and contributor to the startup communities globally. She holds an MBA from the University of British Columbia in Entrepreneurship and Innovation and a Bachelor of Arts from Chulalongkorn University in Thailand.

When not investing, you’d find her scuba diving in the tropical oceans or hiking in the great outdoors - conquering the next mountain summit.

She’s lived in the tech startup world for over 15 years, and has played an active leadership role in the women in tech ecosystem for over a decade, having led a global, 60+ chapter organization focused on women in technology and entrepreneurship as its President for several years.

She’s also an EIR at LaunchNY, and a long-time mentor for Techstars. She was recently CMO of untapt, an AI-driven tech talent marketplace based in NYC and has held the role of Adjunct Professor in PR and Marketing at at Syracuse University.

Driving diversity, equity and inclusion action and awareness has been absolutely central to her work, and she additionally enjoys putting her experience in the tech industry to work advising and mentoring startups.

She holds BAs in Political Science and African American History from the University of Rochester, an MBA from Goizueta Business School at Emory University and an MA in International Relations from the Fletcher School at Tufts University.

Greg began his investment career at Unitus Capital, an investment bank based in Bangalore, India. Before moving into the investment field, Greg worked as a product manager for FINCA International, a global micro-finance bank, and as a grant-writer and researcher at the Amman Center for Human Rights Studies in Amman, Jordan. He also was a SAT/ACT tutor for students in the United States and Jordan. Greg received his B.A. in International Political Economy the University of Illinois at Champaign-Urbana.

Outside of work, Greg enjoys boxing, playing soccer, building things out of wood, and hiking the trails of northern California. 

I am absolutely thrilled to share with everyone that I recently joined the investment team at Scrum Ventures!

I’ve been here in the SF office for a little over a week now and am still pinching myself to make sure that it’s all real. In my new role, I will be investing in, supporting and empowering ambitious founders and globally-focused seed and Series A startups helping to build smart societies for a better future. We are generalist investors, so I will be expanding my focus beyond AgTech to a wide range of technologies, including AI, robotics/automation, AR/VR, and data/cloud/IOT across dynamic industries such as retail, fintech, manufacturing/construction, mobility, sports, health/wellness, and media, among others. Scrum Ventures also has a specific interest in the US and Asia and as strategic investors for startups with cross-border go-to-market interest in both Silicon Valley and Japan. So far, Scrum has invested in 77 companies (including Noom, THIRDLOVE, May Mobility, Le Tote and dozens of other industry defining companies) with an incredible 11 exits to show for it already.

What I love most about Scrum Ventures is that we all value community and embody founder empathy.

This is a team made up of successful serial entrepreneurs and astute startup advisers with incredible foresight and an uncanny ability to use networks and communities to elevate companies. Many of us have been startup operators and founders in a past life, with varying degrees of success. But since we’ve been in the trenches, we know how to help entrepreneurs in more tangible and targeted ways. In addition, with deep networks across both the US and Japan, Scrum is able to seamlessly help companies rapidly expand into some of the most sought-after markets and user populations in the world. But our investment team is only as strong as the community that we build around the companies we choose to align ourselves with. On my first day on the job, we ended the work day by hosting a networking dinner event (with food catered by a Michelin-star chef) for all of our portfolio company founders on the West Coast. The event was incredible and the sense of community and camaraderie among the diverse group of founders present was palpable. Scrum takes care of its entrepreneurs. And this makes me so excited to spend this next chapter of my life standing side by side with Tak, Hitoshi, Yohei, Ryan and the rest of the team in supporting the next generation of transformative entrepreneurs.

For those who are curious, my journey to Scrum started with a seemingly nondescript LinkedIn request.

Nearly half a year ago, I met Scrum’s Ryan Mendoza when he reached out to swap industry knowledge on the AgTech industry and to collaborate on helping to build out Scrum’s thesis in the sector. I remember being really impressed with the hustle and initiative that Ryan showed in conducting market research in an industry completely foreign to him. I also found our conversation particularly dynamic — he’s sharp and analytical, but also genuine and down to earth. And those are the kinds of qualities that I value in both colleagues and friends. Over the next few months, that initial conversation then led to additional chats to share deal flow and to further explore the viability of investing in AgTech, which eventually turned into a collaboration to assist with due diligence on a precision ag company Scrum was interested in investing in. Soon after that, Ryan announced that Scrum was looking to expand their investment team and it was an opportunity that we both saw as a perfect fit. Good fortune and impeccable timing played key roles here.

I’ve doubled down on my commitment to be a VC because I want to support mission-driven entrepreneurs building the kind of world I want to live in.

I stumbled into the world of entrepreneurship in college and started my career with two failed startup experiences. But through those opportunities, I was really able to appreciate the dedication, challenges and sacrifices that entrepreneurs routinely make to solve their consumers’ biggest pain points. When I joined THRIVE, my goal was to advance the future of food and agriculture in order to accelerate a more sustainable future and to drive change in our food production processes to improve the world we live in. Agriculture has long been considered one of the least digitized industries, but several larger and older industries are also ripe for change. According to Exhibit 3 on this graphic published by McKinsey, construction, education, healthcare, hospitality, transportation and retail all fall under the umbrella of industries with significant digitization opportunities. These industries also happen to fall under Scrum’s areas of interest as well as my own. In recent months, I’ve become particularly interested in thinking about how we will introduce automation into legacy industries to offset labor shortages, how we will improve access to healthcare services, how we can make travel safer and more efficient, and how we can implement sustainability and environmentalism into everyday processes. At the end of the day, I have faith that entrepreneurs can help transition us towards a world of abundance, opportunity, impact, and community. I’m fundamentally an optimist and am excited about what the future holds for technology and innovation.

I’m on the hunt for founders who share our vision.

My goal as an investor is to be helpful, humble and empathetic — the same traits I commonly see pervasive among my most favorite entrepreneurs, peers and mentors. If you are a seed or Series A company helping to build smart societies for a better future and have some traction, a differentiated product, world-class team and demonstrated product/market fit, I want to hear from you. Or if you just want feedback on your company or idea, feel free to reach out to me on LinkedIn or share your story with me via email at jonathan@scrum.vc.

Jonathan is an investor and startup mentor dedicated to helping impact-driven entrepreneurs scale and grow their business. Jonathan is regularly invited to host pitch competitions and speak on-stage about his experiences in VC, entrepreneurship, the Silicon Valley startup landscape, and his interest in the AgTech space. In addition, Jonathan has been invited as a speaker at the Forbes Under 30 Summit, was selected as part of the 2019 AgGrad 30 Under 30 class and is a member of the NationSwell Council. Jonathan has a BA in History from Rice University and an MBA from the Johnson School at Cornell University.

Connect With Me: LinkedIn| Medium | ✉

The Next Adventure: I’m Joining Scrum Ventures was originally published in Scrum Ventures on Medium, where people are continuing the conversation by highlighting and responding to this story.