And every time, I see two reactions from founders and from CFO friends.

The first: Finally. Something that will handle this so I don't have to think about it.

The second: AI can't understand the context of my business. A real CFO would know my customers, my seasonality, my deal structure. This is just automation.

Both reactions are understandable. Both are also missing the actual problem.

Because the real problem for most early-stage founders isn't that their CFO doesn't understand their business.

It's that they don't have a CFO.

The research on this is consistent enough that it's basically a rule of thumb: companies typically hire their first CFO somewhere between Series A and $100-250M ARR. Before that threshold, the finance function is assembled on the fly — the CEO handles strategy, a part-time bookkeeper handles transactions, a CPA handles taxes, and whatever gap is left over falls to whoever has time that week.

What that looks like in practice: a bank balance that gets checked more than it should, a cash flow model that's three months stale, a burn rate calculation that doesn't account for the contract that isn't renewing, and a quarterly conversation with your CPA that starts with "so, where are we?"

The gap isn't a CFO gap. It's a visibility gap.

That's the thing AI actually replaces — not the CFO, but the absence where the CFO should be. The thing that gets you from flying completely blind to having a real-time read on your cash position, your burn trajectory, your scenario outcomes.

Will AI make every decision a CFO would make? No. The judgment calls — which expense to cut, when to raise, how to structure a deal, whether to hire — still require a human who understands the business deeply. What AI replaces is the time and friction between "I need to know where we stand" and actually knowing.

I've written before about how cash flow forecasting is a discipline that most founders understand in theory but struggle to maintain in practice. The 13-week rolling forecast isn't complicated. It's just relentless. It needs to be updated. It needs to stay current. It needs to survive the week when payroll just ran, two proposals are due, and you haven't had time to look at the numbers since last Tuesday.

AI doesn't replace the discipline. It makes the discipline sustainable.

The founders who use these tools well aren't the ones who hand the finance function to the AI and walk away. They're the ones who actually read what the AI surfaces, ask the follow-up question, use the time they got back to think harder about the business rather than harder about the spreadsheet.

The bottleneck was never the data entry. It was always the time and attention required to act on what the data was telling you.

If you've been waiting for AI to solve your finance problem, here's what it's actually going to solve: the "I didn't have time to look at this until it was a crisis" problem.

For most founders, that's not a small thing. That's the whole thing.

What would you actually do with an extra hour a week if your financial model was always current?

These are fine questions. They’re not the important ones.

The important question is what actually happens to your company’s financial data — your revenue, your payroll, your margins, your runway — at the moment an AI model processes it. Not in policy. In the architecture.

That question has a specific technical answer. Most products don’t surface it. A few are designing around it intentionally. The difference matters a lot if your customers are businesses with real compliance requirements.

Why this matters more than most people admit

Your company’s financial data is not like your personal data. It’s not your email history or your social media activity.

It’s the information that tells your investors whether to extend your runway. It tells your bank whether to approve your credit line. It tells your competitors — if they ever got it — exactly where you’re vulnerable. Revenue by customer. Payroll by department. Cash position at the end of every month.

When you start moving this data through AI systems, the governance question isn’t theoretical. It’s the thing an enterprise customer’s security team asks before signing. It’s the gap that surfaces in due diligence. It’s what your lawyers should be reviewing before your product goes to market.

Two announcements this month made this concrete: OpenAI launched a personal finance feature connecting to 12,000+ financial institutions, and Anthropic released 10 pre-built finance agent templates. Both good products. Both got reactions that missed the more important question.

The consumer/enterprise confusion

OpenAI’s personal finance feature is a consumer product. This should be obvious, but the reaction to it suggests it isn’t: a meaningful number of business operators read that announcement and started thinking about their company’s finances.

Consumer products and enterprise products are not the same thing. Not in the liability model. Not in the data handling terms. Not in the audit trail. Not in the governance structure.

A consumer product connects to your bank account under consumer terms of service. There’s no data processing agreement governing what happens to your company’s financial data. There’s no dedicated tenant isolation — you’re one of millions of accounts. Training opt-outs may apply, or may not, depending on which plan tier you’re on and what the ToS says this month.

None of this is appropriate for company financial data. Not because OpenAI is careless — they’re not — but because it’s the wrong product category. Connecting your personal checking account to a budgeting AI and connecting your company’s QuickBooks to a financial analytics platform are different decisions with different requirements.

If you’re evaluating AI for your business finances, the product category question comes first. Consumer or enterprise. The technical questions follow from there.

The tool-use architecture question everyone is skipping

Here’s what actually happens when an AI model answers a financial question. Understanding this is prerequisite to understanding the data custody problem.

Modern AI finance tools use a “tool-use” or “function-calling” architecture. The model doesn’t know your company’s finances by default — it has no access to your data at startup. Instead, it reasons about what data it needs and calls tools to retrieve it. The tool queries your database, gets the relevant data, and returns it to the model as context. The model reads that result and generates an answer.

This is a good architecture. It means the model only touches data it actually needs, for the specific query at hand. It’s more efficient and more controlled than dumping your entire transaction history into a prompt.

But here’s the thing that matters for data custody: at the moment that tool result comes back to the model — when your Q1 revenue, your March payroll run, your AR aging — becomes part of the model’s context, that data is in the inference pipeline. Whatever infrastructure is running inference now has access to it.

Where inference happens is therefore the data custody question. Not “will they train on it.” Where does it physically run, under whose infrastructure agreement, and what’s the audit trail.

The spectrum

There’s a real range here, and it determines your compliance story.

Consumer AI products sit at one end. Inference runs on the provider’s consumer infrastructure. Your data is processed under consumer terms. No data processing agreement. No enterprise audit trail.

Direct API sits in the middle. You call an AI provider’s API directly from your application. Your data transits their servers as query context. If you have an enterprise agreement, there’s a data processing agreement and a training opt-out. But inference still happens on their infrastructure, under their terms.

Managed cloud service sits at the other end. Instead of calling an AI provider’s API directly, you invoke the same model through a managed cloud service — AWS Bedrock, for example. Inference happens within that cloud’s infrastructure. The data handling agreement is with the cloud provider, not directly with the model company. You can deploy within your own VPC. The data doesn’t leave your cloud account’s network boundary to reach the model.

Most AI finance tools on the market today are in the middle tier. A few are moving toward the third.

What we built, and what we’re building toward

At MyRunwayHealth, Merlin is our AI financial analyst. We’ve been thinking about this problem since before we wrote the first line of code.

System prompt with zero financial data. Merlin’s system prompt — the instructions that define its behavior — contains no financial information. The model starts every session knowing nothing about your company’s books. This is a deliberate design choice. It means no financial data transits the inference pipeline unless a specific query requires it.

On-demand data retrieval. When you ask Merlin a question, it retrieves only the data needed to answer that specific question, at the moment it’s asked. Not your full transaction history as a baseline. The data minimization principle applied at the architecture layer, not the policy layer.

Tenant isolation enforced in code. Every tool call validates that it’s accessing the right company’s data before executing. Not a config flag. Not a row-level filter. A hard check in the infrastructure layer that returns an access denied response if the tenant doesn’t match. Your data is fully separated from every other company’s data at the database level.

Full audit trail. Every tool call Merlin makes, every piece of data returned, every request to the AI model — logged. The audit trail answers the question “what did the model actually see?” Not just for debugging. Because that’s a real compliance requirement for business financial data.

The Bedrock direction. We’re moving Merlin’s AI inference to AWS Bedrock. Same Claude model we use today. Completely different data custody story.

With direct Anthropic API (where Merlin runs today), your query context — the tool results containing your financial data — transits Anthropic’s infrastructure. Enterprise sub-processor agreement, no training, audited. But Anthropic’s servers.

With Bedrock, inference runs within AWS’s managed infrastructure. The agent runs inside a VPC with private subnets and no inbound internet traffic. Audit logs go to CloudWatch. The data handling agreement is with AWS. Your financial data doesn’t leave your AWS account’s network boundary to reach the model.

That’s not a minor distinction for a business dealing with sensitive financial data. It’s the difference between “we have a data processing agreement with our AI provider” and “your data stays inside your cloud infrastructure.” The second answer is a lot easier to explain to an enterprise customer’s security team. It’s also the right architecture if you’re serious about financial data governance.

The questions worth asking

Before connecting any AI to your business’s financial data — whether you’re buying a product or building one — these are the questions that actually matter:

Is this a consumer product or an enterprise product? The answer determines everything downstream: data handling terms, audit trail, governance structure, liability model.

Where does inference happen? The AI provider’s consumer infrastructure? Their enterprise API? A managed cloud service within your own cloud environment? The answer tells you whose data handling agreement you’re actually operating under.

What data reaches the model, and when? Full transaction history front-loaded? Or only query-relevant data retrieved on demand? This is an architecture question, not a policy question. Ask to see the architecture.

Is there an audit trail? Not “we log things.” A specific answer: every tool call, every piece of data returned to the model, every API request. Queryable. Exportable. Your evidence that the system behaved as designed.

Is tenant isolation enforced in code or in config? Config can be misconfigured. Code-level enforcement at the infrastructure layer cannot.

The AI finance market is moving fast. Most of the energy is going into capabilities: accuracy, speed, integrations, UI. The data governance infrastructure tends to come later — usually when an enterprise deal stalls in security review, or when you realize your ToS doesn’t cover how the product actually works.

Design for it now. Retrofit is expensive.

If you were a Bench customer, you woke up one December morning and discovered you couldn't access your own financial records. Not because you did anything wrong. Because the company running your books ran out of runway.

This spring, Capital One completed its acquisition of Brex for $5.15 billion — less than half what Brex was worth at its peak. When the deal was announced, Brex co-founder Pedro Franceschi published a letter to customers. "Brex will remain Brex," he wrote. Capital One's own internal directive, he noted, was explicit: "Don't crush the butterfly."

I applaud his intent. History has told us a different story.

These aren't unrelated events. They're the same story playing out across every layer of the founder finance stack.

The fintech tools that emerged over the past decade to serve founders and small business owners are consolidating. Much of this is being driven by the suddenly new unit-economics reshaping brought on by AI (we'll talk more about this next week! Be sure to subscribe so you don't miss out…)

Subscribe now

The ones that survive this AI-induced reframe will likely do so by scaling up market rather than by completely restructuring their business — by chasing enterprise customers, by integrating into larger platforms, by building for the $50M ARR company, not the $500K one.

Which means the window is narrowing for founders who've built their financial operations on tools they thought were designed for them.

And it's not just the transactional layer. It's the planning layer too.

Causal — one of the cleaner FP&A and financial modeling tools built specifically for startups — was acquired by LucaNet in late 2024. LucaNet is a serious enterprise finance platform, which is exactly the problem: Causal gets folded into a product roadmap now oriented around corporate consolidation and ESG reporting, not the scrappy founder trying to model out a three-scenario cash runway.

Mosaic, which had carved out a real following among Series A and B companies for collaborative FP&A, was acquired by HiBob in early 2025 for around $35 million. HiBob is an HR platform. Not a finance platform. The FP&A tooling got absorbed into a people management product because someone decided the integration story made sense — for them. Whether it still makes sense for the founders who built their finance workflows around it is a different question.

The pattern isn't hard to see. Bookkeeping tools get acquired by banks. FP&A tools get absorbed by enterprise platforms or HR software. Expense management goes to the card issuers. Each acquisition makes a certain kind of strategic sense for the acquirer. None of them are optimizing for the $2M ARR founder or SMB owner who just needs a reliable tool that does the thing it says it does.

Here's what I've seen happen when a finance tool gets acquired or shuts down:

The data migration is always messier than anyone tells you. Months of bookkeeping history gets lost in the handoff. Your audit history suddenly has gaps that you don't know if you can rebuild. The reconciliation you thought was done turns out to have been stored in a proprietary format, and now your new tool can't read it. You end up rebuilding from bank statements.

It's not catastrophic. But it costs time you didn't have and weren't ready to allocate.

And that's the optimistic version — the one where the acquiring company actually keeps the product alive. Apple acquired Dark Sky in 2020, and by end of 2022 it was gone, with the technology folded into Apple's own Weather app. The standalone product itself quietly shut down. Dark Sky didn't fail. It got acquired.

What really changes when a tool gets acquired isn't the logo. It's the roadmap. New ownership means new priorities. Features that were free become premium. Support tiers that matched your budget disappear. The thing you built your workflow around stops working the way it used to.

I've written before about how your bank balance isn't actually telling you your real financial position. This is the same problem, one layer up: the tool isn't always going to be there, either.

I'm not saying avoid all fintech tools. That's not practical advice, and it's not what most should do - The 'build vs buy' argument has also changed.

What I am saying: build a little more redundancy than you think you need.

Your financial records should live somewhere you can always access them, regardless of what happens to the tool that generated them. Your monthly exports should be a habit, not a crisis recovery plan. If the FP&A tool you rely on for scenario modeling disappeared tomorrow, could you recreate your last 90-day forecast without it?

If the answer is no, that's a risk worth knowing about.

The founders I've seen handle this well don't obsess over picking the perfect tool. They pick good tools and build independent records alongside them. The tool is the shortcut. Your actual numbers — the real data about your business — are the source of truth. Those need to live somewhere you own, not just in a dashboard you're renting.

Runway calculators, scenario models, cash flow forecasts — whatever you're using to understand your financial position, that understanding needs to be portable (don't mistake this for 'ex-portable'!). Not locked to a platform that's one acquisition away from becoming someone else's enterprise upsell.

The consolidation is going to continue. The market is picking winners, and they're mostly not picking you. In five years, the finance stack for founders will probably look very different from what it looks like today.

That's not a reason to panic. It's a reason to hold your tools loosely and your data tightly.

What's your backup plan if your primary financial tool disappeared tomorrow?

It is the right question. But it is also the wrong framing.

The real question is not how to get clients to pay faster. It is why your accounts receivable process is designed to let them pay slowly in the first place.

The Silent Revenue Leak

Here is a number that will ruin your morning: the average small business has 24% of its total revenue sitting in unpaid invoices at any given time. For B2B service companies, it is closer to 35%.

Think about that. A third of the money you have earned — work you have already done, value you have already delivered — is sitting in someone else’s bank account, earning them interest instead of you.

We talked about this dynamic in the profitable-not-fine post. Profit means nothing if the cash is stuck in AR. But what I did not get into last week is just how much that stuck cash actually costs you.

The Compound Cost Nobody Calculates

Late payments are not just an inconvenience. They have a real, calculable cost that most founders never bother to quantify.

Let us say you have $150K in Accounts Receivable (AR) with an average collection period of 52 days. If you could bring that down to 30 days, you would free up roughly $55K in working capital. That is not theoretical money. That is cash you could use to hire, invest, or simply stop worrying about making payroll.

But it gets worse. That $55K sitting in limbo forces you into other expensive decisions. Maybe you take on a line of credit at 12% to cover the gap. Maybe you delay hiring a key role by two months. Maybe you take a bad deal because you need the cash now rather than the right cash later.

The cost of slow AR is never just the float. It is every downstream decision you make because the float exists.

Why Smart Founders Have Dumb AR Processes

I have worked with hundreds of startups at this point. The pattern is almost universal: brilliant product founders with absolutely no accounts receivable process.

Early on, invoices go out whenever someone remembers. Payment terms are whatever the client asked for during the sales process. Follow-up on late payments is awkward and inconsistent because the person sending the invoice is often the same person managing the client relationship.

This is not a character flaw. It is a structural problem. When you are a ten-person company and the CEO is also the head of sales and the de facto collections department, something is going to give. Usually it’s a routine process that “It’s okay if I do this next week instead”. And that’s often your cashflow review.

In other words, AR.

Five Things That Actually Move the Needle

I am not going to pretend there is a magic fix. But after watching what works across a few hundred companies, these are the five changes that consistently make the biggest difference:

Shorten your default terms. If you are offering Net 60 because that is what your first client asked for three years ago, stop. Net 30 should be your standard. Net 15 is better if you can get it. Every day you shave off your payment terms is a day of cash you get back.

Invoice the same day you deliver. Not next week. Not at the end of the month. The day the work is done. The longer the gap between delivery and invoice, the longer the gap between invoice and payment. Psychology matters here — when the value is fresh, the urgency to pay is higher.

Automate your follow-up. This is the single biggest unlock for most small companies. A polite, automated reminder at 7 days, 14 days, and 21 days past due is not aggressive. It is professional. And it removes the awkwardness of the founder having to chase payments personally.

Offer a 2% discount for Net 10. This sounds like you are giving away money. You are not. You are buying cash flow certainty. 2% to get paid 20 or 50 days early is one of the best deals in business finance. Do the math on your cost of capital and you will see it immediately.

Review AR aging weekly. Not monthly. Weekly. It takes five minutes. Look at what is overdue, what is coming due, and what your total AR balance looks like. This one habit prevents more cash crises than any financial tool I have ever seen.

The Connection to Everything Else

AR does not exist in isolation. It connects directly to your burn rate, your cash runway, and your ability to grow without external funding. When your AR process is broken, everything downstream suffers.

At MyRunwayHealth, we don't fix your AR process — that is not our lane. But we do live in your cash flow forecast all week long, which is exactly where slow-paying invoices show up first. AR aging is rarely the most exciting page in your books, but it is often the fastest path to improving cash position without changing anything else about the business. (If you want to actually overhaul the operations side, Duncan at Lunos.ai — who joined me on the March 31 webinar — is who we send people to.)

You do not need to raise money. You do not need to cut costs. You just need to collect the money you have already earned.

When was the last time you actually looked at your AR aging report? If the answer is “what is an AR aging report,” start with Duncan — and we will keep an eye on your cash flow from ours side.

I have seen this movie more times than I can count. A founder calls me, confused and a little scared. “We had our best quarter ever. Revenue is up 40%. But I’m not sure we can make payroll next month without a line of credit.” They are not bad at math. They are experiencing one of the most counterintuitive realities in business: profit is an opinion, but cash is a fact.

The Paper Profit Trap

Your P&L says you made $80K last quarter. Great. But that number includes $120K in outstanding invoices that have not been paid yet, minus expenses you already covered out of pocket. The P&L does not care when money moves. It cares when revenue is recognized and expenses are incurred. Those are accounting concepts, not bank balances.

This isn’t a bug. Accrual accounting exists for good reasons. But if you are running a growing company and treating your income statement like a bank statement, you are going to have a bad time.

Growth Eats Cash for Breakfast

Here is what makes this truly maddening: the faster you grow, the worse it gets. Growth requires investment — hiring ahead of revenue, buying inventory before sales close, paying for infrastructure before customers onboard. Every dollar of growth creates a cash gap between when you spend and when you collect.

I worked with a founder a few years ago — let’s call her Sarah — who had $200K in accounts receivable, a 67-day average payment cycle, and $85K in monthly burn. On paper, she was killing it. In practice, she was three slow-paying clients away from a crisis.

The Three Numbers That Actually Matter

Forget your P&L for a minute. If you want to know whether your business is actually fine, look at three things:

Days Sales Outstanding (DSO). How long does it take to collect after you invoice? If this number is going up, your profit is becoming increasingly fictional.

Operating Cash Flow. Not net income. Not EBITDA. Actual cash generated (or consumed) by operations. This is the number that determines whether you can pay people.

Cash Runway. At your current burn rate, how many months of cash do you have? If the answer is less than three, it does not matter what your P&L says.Why Your Monthly P&L Is Lying to You

Okay… “Lying” is strong. Let us say “omitting critical context.” Your monthly P&L does not show you the timing of cash flows. It does not show you that your biggest client seems to have switched to Net 90 terms. It does not show you that your Q2 insurance premium and annual software renewals all hit in the same month.

A monthly P&L is a rearview mirror. Cash flow forecasting is the windshield. You need both, but if you can only look at one, look through the windshield.

This is exactly the kind of thing we talked about with burn rate math a couple of weeks ago — it is all connected. Revenue recognition, burn rate, cash position. You cannot understand any one of them in isolation.

The Fix Is Boring (And That Is Why It Works)

The solution is not complicated. It is a 13-week cash flow forecast. Every Monday, you update it. You look at what is coming in, what is going out, and what the gap looks like. It takes 30 minutes once you have the template set up.

The founders who do this sleep better. Not because the numbers are always good, but because they are never surprised. And in the early stages of a company, surprises are what kill you.

At Runway Health, this is one of the key modules we’re building. Not because it is glamorous — it can literally be a spreadsheet (don’t worry, ours won’t be!) — but because it is the single highest-leverage financial tool a founder can use. It turns “I think we are fine” into “I know exactly where we stand.”

Profitable but not fine. It is more common than you think. What is your DSO right now — do you actually know?

It’s a sharp piece. Worth reading. And the advice is sound if you’re running a venture-backed SaaS company with a board that’s benchmarking you against the Bessemer Cloud Index. But the part that stuck with me wasn’t the two paths. It was what sits underneath them.

The Question Nobody Wants to Update

Every company — startup, small business, nonprofit — is built on a set of assumptions. How your customers buy. What they’ll pay. What it costs you to deliver. What “good enough” looks like in your market. Those assumptions were true when you made them. The question George is really asking is whether they’re still true now.

And it’s worth being honest about this: a lot of them probably aren’t.

Per-seat pricing — the foundation of basically every SaaS company built in the last fifteen years — is under real pressure. When one person using AI tools can do the work that used to require three seats, the math that underpinned your revenue model starts to wobble. Chargebee’s research found that 37% of SaaS companies are planning pricing model changes in the next twelve months. I can confirm that with conversations that are happening within the Operators Guild, where founders and operators are sharing the same pattern. That’s not a trend. That’s a reckoning.

Meanwhile, AI-first products are running at roughly 52% gross margins — compared to the 75-85% that traditional SaaS companies got comfortable with. The cost structure has shifted underneath the business model, and a lot of founders haven’t fully reckoned with what that means for their unit economics.

This isn’t abstract. This is the spreadsheet or model you’re not updating.What If Everything You Built For Changes?

Here’s the question that keeps showing up in my conversations with founders and operators:

What if the way our customers buy fundamentally changes in the next eighteen months?

Actual output from Merlin, the AI financial engine inside MyRunwayHealth.

That’s not a hypothetical. Buying processes are already being reshaped by AI. Procurement teams are using AI to evaluate vendors. Experienced operators are comparing your product against what they can build themselves with off-the-shelf AI tools. Decision-making timelines are compressing — or, in some cases, evaporating entirely because the buyer just built a “good enough for now” version over a weekend.

George’s framework gives you two doors. But before you pick one, you need to answer a more basic question: does the business model you’re running still reflect the market you’re in?

The Board Problem (and the Investor Blind Spot)

This is where it gets uncomfortable for companies that have raised capital.

The VC ecosystem doesn’t just fund companies. It shapes them. The playbook that comes with institutional money — hire fast, capture market share, optimize unit economics later — was designed for a world where software moats were durable and switching costs were high. That world is eroding.

And the board dynamics that come with that playbook can make it harder, not easier, to adapt. When your investors are benchmarking you against a model that assumes 18-month sales cycles and enterprise stickiness, it’s difficult to raise your hand and say, “I think we need to rethink the whole go-to-market.” Even when that’s exactly what needs to happen.

The companies that navigate this well will be the ones where the board and the founders are asking the same hard questions — not the ones where the board is pushing for growth metrics that no longer map to how the market actually works.Do You Still Need a CTO? (A Controversial Sidebar)

One of the long-standing truths of SaaS startups — especially in the eyes of investors — is that you need a technical co-founder. A CTO. Someone who can build the thing.

I think that assumption is worth revisiting too.

Don’t misunderstand — technical depth still matters. But what “technical” means is changing fast. A founding team with strong product instincts, a clear vision for the user, and the ability to leverage AI development tools can now move at a speed that would have required a 10-person engineering team three years ago. The question used to be “who is writing the code” — when now it may be more of who is asking the right prompts, who understands the user deeply enough to direct the build, and who can iterate on a living product without a six-month development cycle.

There is a striking example of this from a completely different field. A Harvard physics professor recently used Claude — the same AI tool I use to build parts of our own product — to compress what would normally be a one-to-two-year graduate-student-level research project into about two weeks. His takeaway wasn’t that AI replaced expertise. It was that domain knowledge and judgment — knowing what questions to ask, and catching when the AI gets it wrong — mattered more than ever. The “technical skill” just shifted from doing the work yourself to directing the work and verifying the output.

That maps directly to what I’m seeing in startups. The founders I’d watch aren’t necessarily the ones with the deepest technical pedigree. They are the ones who understand their customer so completely that they can use every tool available — AI-native or otherwise — to close the gap between insight and product. That’s a different kind of technical. And it might be the kind that matters most right now.So What Do You Actually Do?

If George’s piece made you uncomfortable, good. That discomfort is useful. Here’s what I’d suggest:

Go back to your business model. Not your P&L — your actual model. The assumptions about who buys, why they buy, what they pay, and what it costs you to deliver. Write them down. Then ask yourself which of those assumptions were validated more than six months ago and have not been tested since.

If your answer is “most of them” — that’s your starting point.

The companies that come out of this period strongest won’t be the ones that picked the right door from George’s framework. They’ll be the ones that had the discipline to question whether the building they are standing in is still structurally sound.

When’s the last time your company actually interrogated its own business model — not the revenue line, but the assumptions underneath it? I’d love to hear what you’re seeing. Drop a comment or hit me up directly.

Related: There Are Only Two Paths Left for Software — David George, a16z

Reference: Vibe Physics (Anthropic Research) — A Harvard physics professor used Claude to compress a 1-2 year grad-student-level research project into 2 weeks.

A few of you replied with some version of: Okay, I’m terrified now. What do I do about it?

Fair. Let’s talk about the tool that answers that question — and no, it’s not a more complicated spreadsheet.

Scenario planning sounds like something a Fortune 500 CFO does in a boardroom with a 47-tab Excel model and a team of analysts. And honestly? That’s not entirely wrong. Big companies do this religiously. They map out best case, worst case, and base case. They stress-test assumptions. They run the numbers before the numbers run them.

But the concept itself is dead simple: What happens if things change?

What happens if your biggest client leaves? If you hire two people next quarter? If revenue dips 20% for three months? If that grant you’re counting on doesn’t come through?

These aren’t paranoid questions. They’re survival questions. And right now, they’re more relevant than ever.

If you’re a nonprofit, you already know what I’m about to say. Federal funding is being cut mid-cycle across thousands of programs, with non-defense spending facing a 22% reduction in the proposed FY2026 budget. Grants that organizations planned around — budgeted around, hired around — are disappearing. And it’s not just federal. The ripple effect hits state funding, foundation priorities, and donor confidence.

If you’re a bootstrapped startup, you’re facing a different version of the same problem. Capital is scarce. The customers you’re counting on might churn. That partnership might not close. The economy might do something weird in Q3.

The point isn’t to predict the future. It’s to stop being surprised by it.

So here’s what scenario planning actually looks like when you strip away the MBA jargon:

You take your current financial picture — your revenue, your expenses, your cash position — and you ask three questions. What if things go well? What if things stay flat? What if something breaks?

For each scenario, you adjust the inputs. Revenue up 15%. Revenue flat. Revenue down 20%. Then you look at where each path takes your cash position over the next three, six, twelve months.

In theory, this takes about an hour.

In practice? You need current numbers (not last month’s — this week’s). You need to model each variable independently. You need to remember that expenses aren’t static — they shift with headcount, seasonality, vendor contracts, and the forty-seven subscription tools you forgot you’re paying for. And you need to do this regularly, because the moment your assumptions go stale, the whole exercise becomes fiction.

Most founders I talk to have tried this once. Built a three-tab spreadsheet on a Sunday. Felt great about it. Opened is twice. Never opened it again.

The spreadsheet isn’t the problem. The maintenance is.

This is the gap we built MyRunwayHealth to close. Not the math — founders can do math. The consistency. For me, it was the time it took to go find the updated numbers to plug back in to the worksheet, and then that extra 40 minutes trying to figure out where I clicked to break that one formula. The part where you need current bank data, accurate expense tracking, and the ability to toggle assumptions and immediately see what happens to your runway. Without spending your Saturday rebuilding formulas.

A scenario plan you built two months ago is just a story you told yourself. A live one — one that updates when your numbers change and shows you the impact of decisions before you make them — that’s how you stop being blindsided.

The best practice isn’t having a plan for when things go wrong. It’s having a plan you actually keep current.

What’s the one scenario that would change everything for your business? And have you

actually modeled what it would look like? I’d love to hear what keeps you up at night.

Subscribe now

For me, it was spreadsheets. I spent years watching early-stage founders, and a number of finance people — all smart, capable people — try to manage their financial future in a tool built for static data entry. They’d build a beautiful cash flow model on a Sunday night, and by Wednesday it was already subtly lying to them. Not because the math was wrong, but because the world had moved and the spreadsheet hadn’t.

That’s why we built MyRunwayHealth. Not because spreadsheets are bad — they’re incredible, actually — but because using one to manage a living, breathing cash position is like using a paper map to navigate through a construction zone during rush hour traffic. The tool wasn’t built for the job.

I’ve been thinking about this idea lately — the gap between “the tool everyone uses” and “the tool that actually works” — because I recently met someone who’s fighting the exact same battle in a completely different arena.

Meet Neima

“After a decade in corporate finance, I got very good at quantifying/predicting the downstream financial results of decisions. Simultaneously, I was so sick of applying for jobs and having my resume shredded by the screener, I felt like there had to be a better way—so I built it.”

Neima Beizai is the founder of Your Opportunity Co., and he has a theory about resumes that I think anyone who’s ever hired someone — or been hired — will immediately recognize.

Here it is: resumes have become the spreadsheets of hiring.

Think about it. Resumes were designed decades ago to give employers a quick snapshot of a candidate. Name, experience, education, skills. It worked fine when hiring was local, applicant pools were manageable, and people stayed in the same career track for twenty years.

That world is gone. And the resume didn’t evolve with it.

The Arms Race Nobody’s Winning

What’s happening right now in hiring is almost comically dysfunctional. Candidates are using AI to write resumes optimized for keyword filters. Companies are using AI to screen those AI-written resumes. It’s an optimization loop with no human in it — two algorithms trying to out-game each other while the actual person behind the application becomes increasingly invisible.

Jamie Ceglarz, who runs Guild Talent, the recruiting arm of the Operators Guild, posted recently about a Head of Finance search that got 1,000 applicants on the first day — they had to take the posting down. His take: technology → more volume → more noise → more need for signal. That’s not a recruiter complaining about competition. That’s someone watching the system buckle under its own weight.

“AI-on-AI violence is absurd to the highest degree. I’ve had so many conversations with talent teams where someone got to interview who was clearly using AI during the interview, or even worse, they were hired only for the organization to discover they couldn’t deliver. All the while, AI interview tools are being used to police applicants and try to catch them to filter them out, resulting in increased frustration and drop off, meaning there is clearly talent and money being left on the table. No one is winning, everyone in suffering.”

And the old problems haven’t gone away — they’ve gotten worse. Embellished credentials. Keyword stuffing. The quiet reality that some of the best people for a role are terrible at writing about themselves on paper. The resume was already a lousy way to capture what makes someone good at a job. Now it’s a lousy summary written by a chatbot, filtered by another chatbot, and reviewed by a hiring manager who has 90 seconds and a headache.

Sound familiar? It should. It’s the same pattern we see in finance every day — a legacy tool being pushed way past its design limits, with everyone just kind of accepting that “this is how it works” because they’ve never seen an alternative.

What If You Changed the Dimension?

Here’s where Neima’s approach gets interesting, and where I think it connects to something bigger.

Instead of trying to build a better resume filter, Your Opportunity Co. changed the evaluation dimension entirely. They said: what if, instead of reading about what someone claims they can do, you actually watched them demonstrate it?

YOC empowers you to build customizable, role-specific Smart Assessments. When a candidate applies, they get an optional link to complete an assessment tailored to that specific role. Not a generic personality quiz — a real evaluation of the thinking, problem-solving, and skills the job actually requires.

When they complete it, the hiring manager doesn’t get a stack of resumes to sift through. They get a shortlist of qualified candidates, each with a detailed profile: strengths, areas for development, onboarding suggestions, interview focus areas, and the reasoning behind every recommendation.

“In practice, we’re seeing teams save time and improve pipeline quality through reducing decision fatigue from having a clear read on candidates upfront. We have also seen expansion in the candidate pool while simultaneously filtering it down. One story really stands out—a team was hiring for a role that required a high degree of empathy. An individual was highly recommended by our system however their resume didn’t tell the story they would be able to do the role, which means traditionally they’d be filtered out. When the recruiter looked at the Smart Assessment, they instantly knew the candidate was a fit and called them in for an interview.”

The part I love most is that the assessment is optional. That’s not a bug — it’s a feature. A candidate choosing to opt in is itself a signal. It says: I’m willing to show you what I can do, not just tell you about it. That self-selection alone changes the quality of the pool.

The Pattern

I keep coming back to this parallel because I think it reveals something worth naming.

In finance, we had a tool (the spreadsheet) that was “good enough” for decades, until the speed and complexity of modern business broke it. The answer wasn’t a better spreadsheet — it was a fundamentally different approach to keeping business owners connected to their financial reality.

In hiring, they have a tool (the resume) that was “good enough” for decades, until AI and the modern job market broke it. The answer isn’t a better resume filter — it’s a fundamentally different approach to understanding whether a candidate can actually do the job.

Both problems share the same root: we got so used to the legacy tool that we forgot to ask whether it was still the right one.

“During my corporate finance career, I rose to the rank of Head of FP&A and saw up close how much a static tool like a spreadsheet could create disasters I didn’t even expect. When a tool is designed to be heavily reliant on user input, it also requires the user to have the informational awareness to keep it up to date. Rob said earlier that you make a gorgeous spreadsheet Sunday night and by Wednesday it’s all wrong— I felt that at my core, the literal story of my past life. Legacy tools, while useful, were made for a world that no longer exists yet are expected to still be used for things they are not designed for. Informational awareness can now happen automatically and constantly, which is where MyRunwayhealth comes in. I love that they’re fighting the good fight to make forecasting more dynamic and informationally aware. This is game changing since I can check assumptions rather than chase them down, which is its own kind of special torment. It allows me to stay focused on the validation and decision-making element of the forecast, rather than the maintenance of it.”

What This Means for You

If you’re running a small business or a startup, you’re probably dealing with both of these problems right now. You’re managing your cash flow in a tool that can’t keep up. And the next time you need to hire someone, you’re going to drown in AI-generated resumes that all sound the same.

The good news is you don’t have to accept either one as inevitable.

For the finance side, that’s what we’re building at MyRunwayHealth — an always-current view of your cash position, runway, and financial health that doesn’t depend on you manually updating a spreadsheet every week.

For the hiring side, go talk to Neima. Seriously. What Your Opportunity Co. is building is the kind of thing that makes you wonder why everyone isn’t doing it this way already.

“I care deeply about increasing access to opportunity. By replacing the resume screen, we level the playing field, humanize hiring all while making opportunity more accessible.”

The best tools don’t ask you to work harder. They change the game so the hard work actually matters.

Have you dealt with either of these problems — the stale spreadsheet or the resume black hole? I’d love to hear which one keeps you up at night.

— Rob

Subscribe now

Rob Owen is the CEO and Co-Founder of Alidade Systems, the company behind MyRunwayHealth — the AI-native financial intelligence platform for small businesses, startups, and non-profits. He writes the Manual Mode newsletter every week.

Neima Beizai is the Founder and CEO of Your Opportunity Co. — a hiring intelligence platform that replaces the resume screen to help organizations connect with qualified candidates through customizable, role-specific Smart Assessments.

But here’s the thing a forecast can’t tell you if you’re feeding it the wrong inputs. And most founders are.

Let’s talk about runway — specifically, why yours is almost certainly shorter than the number in your head.

Read more

That version exists. But it’s not the one that gets most people.

The one that gets most people is quieter. It’s the slow drift — the weeks where y…

Read more

A few of you replied with some version of the same question: Okay, so what do I actually do about it?

Fair. Let’s talk about the tool that…

Read more

The moment the spreadsheet breaks.

Not literally — though that happens too, usually with a cascade of #REF errors and a formula that worked perfectly fine until it didn't. But the real break isn't technical. It's the moment you look at the spreadsheet you've spent two years building and realize, somewhere in the back of your mind, …

Read more

You check the account. There's a solid number sitting there — more than you expected, maybe. For one brief second, everything feels manageable. You breathe a little.

And then your brain catches up.

Wait. That invoice from March still hasn't come in. And the supplier bill is due Friday. And didn't we agree to…

Read more

The founder might say: "Burn is under control, we've got 14 months of runway, and we're starting to see early traction on enterprise."

The bus…

Read more

It usually happens late at night — or first thing in the morning, depending on your relationship with sleep. You've got a few employees. Revenue is coming in. Things are working. And then a question floats across your brain that you're not quite ready to answer:

Where is all the money going?

…

Read more

Nobody warned you that “running a business” would mean also being the accountant, the HR department, the IT helpdesk, the marketing team, and — on particularly special days — the person unclogging the office sink.

Welcome to the club. It’s crowded in here.

Manual Mode is a newsletter for small business owners a…

Read more