What is an AI wrapper?

It is a product that is mostly a thin layer over a model someone else runs. That is our working definition, not a term either source defines. Troutman puts the problem in one line: "If a competitor can recreate your company or your AI product quickly using publicly available tools, you lack strong defensibility."

Are AI wrappers dead?

Neither source says so. ICONIQ's 2026 survey of over 300 executives at software companies building AI products found that close to two-thirds are working on horizontal or vertical AI applications. Building on the application layer is the common choice, so it is not the problem.

The test is what sits around the model. Troutman writes: "AI defensibility is not about the model, it's about what surrounds the model: data, users, workflows, and positioning." It names four places a moat can come from:

1. Proprietary data and algorithms: exclusive data, properly obtained.
2. Operational dependence and stickiness: AI embedded in customer workflows, so switching costs build.
3. Scale, distribution and network effects.
4. Industry knowledge and regulatory expertise, for example in healthcare, finance and data privacy.

ICONIQ points the same way on products. It says a product embedded in a real workflow appears to be harder to dislodge than a general-purpose model one layer up.

Are AI wrappers profitable?

We found no 2026 source that measures wrappers separately. The nearest primary data is ICONIQ's survey. It reports gross margins on AI products of 45% in 2025, a projected 53% in 2026 and 59% in 2027, and says two-thirds of companies report better per-query unit economics.

It also reports that companies run an average of about 3.3 models, with licensed third-party APIs the most used type. And it flags a cost surprise: one builder watched an agent workflow budgeted at ten cents a run drift past a dollar fifty as agents retried and corrected themselves.

These are survey averages across software companies, not a benchmark for your product. An investor will want your gross margin after model and inference costs, month by month.

What paper do investors read about your AI?

Troutman says AI-specific diligence and representations are rapidly becoming standard in deal documents. Among the representations it lists:

• Disclosure of all third-party foundation models, LLMs, model APIs and fine-tuning services used in your products, and compliance with their terms of use.
• That no such usage gives the provider ownership of your prompts, outputs, embeddings or confidential information.
• That your AI inputs were obtained and used in line with the applicable terms and laws.
• That written claims about what the AI can do are true and supported, with known limitations disclosed.

The article also says patents are hard to get for generic use of off-the-shelf models, and that trade secrets are where "the real moat resides". Its warning is that AI products without trade secrets or truly novel inventions are weakly defensible from a pure IP standpoint.

What does an investor want ready?

Put the two sources together and the diligence list has four items:

1. Which model providers the product depends on, and on what terms.
2. Where the product's own data comes from, and the right to use it.
3. How the product sits inside the customer's work: usage, retention, and how long a customer would take to switch.
4. Gross margin after model costs, and its trend.

The numbers an investor reads next are in The ten numbers an investor reads first. For the cash side, see how much runway a startup needs when it already has revenue.

Work through your answer with us

If a founder wants to say the answers out loud before an investor asks, the Brutally Honest Mentor AI is an AI voice mentor we built and trained on Ozan Sonmez's method. The first 20 minutes are free.

TBHM Growth, our four-week online programme for founders whose company has revenue, works through gross margin, burn and runway in its third week. The next cohort is announced on the Growth page.

Why the 18-month rule does not fit a company with revenue

The rule comes from the day a round closes. J.P. Morgan's guide for startups puts it plainly: "For many years, the rule of thumb was to raise enough capital to last 18 to 24 months." The same guide says that in a tighter market a runway of 24 to 36 months is now recommended more often, and that investors look more cautiously at startups with less than six months of runway.

Read that last point with six to eight months in the bank. You're close to the line investors read with extra care, and a raise you start today will probably close below it.

The calendar between rounds has also stretched. Carta's review of 2025, published in February 2026, found that the median startup closing a Series A had waited about 2.1 years since its seed round, up from 1.5 years five years earlier. Between Series A and Series B the median gap was 2.5 years, against 1.8 years in 2020.

The same report counted 29% of seed and Series A deals on Carta in 2025 as bridge rounds, which Carta defines as any round raised after the first one in a series. A bridge is a planning tool, and it only helps when you know what it has to buy.

So the question worth answering is what your months will do.

Step 1: Find out whether you are default alive

Paul Graham asked the question in 2015, in an essay that still works because it needs only numbers you already have: "Assuming their expenses remain constant and their revenue growth is what it has been over the last several months, do they make it to profitability on the money they have left?" If yes, the company is default alive. If no, it is default dead. Graham's warning in the same essay is that founders assume raising more money will be easy, and the more they depend on that, the less true it becomes.

Fill in four lines, with every co-founder in the room:

Simple runway is cash divided by net burn: 600,000 divided by 70,000 is about 8.6 months. That number hides the thing that decides your next year, because revenue is moving and the burn moves with it. So run the same four lines month by month, letting revenue grow and taking each month's gap out of the cash.

In the example, the answer turns on two dials, growth and costs:

• At 4% monthly growth, revenue never catches the 130,000 of costs in time. The cash runs out during month 11. The company is default dead.
• At 6% monthly growth, revenue passes costs in month 15 with about 41,000 EUR left. Default alive, with almost no margin.
• At 4% growth with costs cut to 115,000 a month, revenue passes costs in month 18 with about 67,000 EUR left. Default alive, and the cut is the reason.

The figures are an illustration, rounded to the nearest thousand, so use your own. If you want the same test explained a second way, Mercury's blog explains it.

Step 2: Write plan B and put a date on it

Graham's advice from the same essay is the part founders skip: "you should know (as in write down) precisely what you'll need to do to survive if you can't raise more money, and precisely when you'll have to switch to plan B if plan A isn't working."

On the worksheet, plan B for a company with revenue is the cost line. Which costs go, in what order, and what the revenue line needs to look like for the company to become default alive without new money. The date matters as much as the list. It's the day the cash left equals what plan B needs to work, and it should be in the calendar of every co-founder.

Before you agree on the date, ask each co-founder separately how many months of cash the company has and what happens on that day. If the answers differ, that's the first conversation to have, and it's better had before an investor asks the same question. We built our startup assessment on the same idea: each co-founder answers on their own, and the report sets the answers side by side, so you see where you agree and where you do not.

Step 3: Count the raise backwards from the money

Raising takes longer than the meetings. Graham's line from 2013 holds for any founder with a product to run: "When you start fundraising, everything else grinds to a halt." Plan for that cost as well as for the time.

Work backwards from the day you need money in the bank:

1. Take your plan B date from step 2. That is the latest day the money can arrive without plan B starting.
2. Subtract the time you expect from first meeting to money in the bank. Not sure how long that is? Ask two investors you trust how long their last deals took from first call to wire.
3. Check the months in between against the calendar. In our experience, first meetings with European investors are hard to get from late November to the first week of January, so a process that starts in that window waits for January. We wrote about planning a raise around that gap in Founders Shift.
4. If the answer lands before today, start plan B now and raise alongside it.

When you do start, run the meetings in parallel. Graham's reason in the same essay is the one that matters on a short clock: "if you only talk to one investor at a time, they don't have the pressure of other investors to make them act."

The order the worksheet follows

With six to eight months and revenue, the worksheet runs in this order:

1. This week, run the default alive test on real numbers with every co-founder present.
2. Write plan B and its date, and agree on both.
3. If the company is default alive, decide whether to raise at all. And if you do raise, you raise from strength, so you can walk away from a bad term sheet.
4. If it is default dead, plan B starts on its date whatever the investor meetings say. A bridge from your existing investors can buy time, so size it to cover the result the next lead investor will pay for and the raise that follows.

What the company does with those months matters more than the number on the runway line.

Work through it with us

TBHM Growth is a four-week programme for founders whose company has revenue, paying customers and six to eight months of runway. Ten startups per cohort spend the four weeks deciding where the company goes next: its position, the founding team, the fundamentals of fundraising and the route. It costs 1,200 EUR per company, co-founders included, and it takes no equity. See if TBHM Growth fits your company.

If you want to talk your own numbers through first, the Brutally Honest Mentor AI gives you 20 minutes free.

Why the order an investor reads in matters

In our experience the decision starts before the meeting. An investor who has read your deck already knows whether you fit their mandate and their cheque size, and an experienced one arrives with a checklist in their head. We think the first three minutes of the meeting decide more than founders expect.

Andreessen Horowitz, the investment firm, set out a reading order in 2015, in its own guide to startup metrics. Investors, the authors wrote, often looked at revenue and bookings first, to gauge the size of the business, then at growth, and those basics, if interesting, made them look further. They added that a founder can present the numbers in any order. So the order is the investor's, and knowing it tells you which numbers have to be ready, and checked, before the first call.

The same guide also warned that people may use different definitions of the same metric, which makes the health of a business hard to read. Lifetime value was one of the numbers they flagged. Built on revenue or gross margin, it suggests a higher limit on what you can spend to win a customer than when it is built on the profit a customer actually contributes. Write the formula next to every number you show, and use the same one every month.

When we read a company's numbers, we look for founders who can tell the company's story with them, without an accountant, a full-time CFO or a live dashboard, and who have a thoughtful answer for each number, including the ones they do not have yet. Read in order, the ten numbers below are that story.

The ten numbers, in the order we read them

These ten come from the money questions in our startup assessment, which each co-founder answers alone. We read them in this order because each one leans on the one before: what comes in, whether it stays, who pays, what a customer costs to win, what the company burns, how long the cash lasts and what the next round has to buy. With revenue, you can work out every one of them from your own records and plans.

The examples use one illustrative company: 60,000 EUR of revenue last month, 130,000 EUR of spending and 600,000 EUR in the bank.

1. Revenue last month

The company's total revenue in the last full month, with the recurring part shown apart. Stripe's guide to SaaS metrics calculates ARR from recurring revenue only, leaving out one-off payments and professional services. In the example: 60,000 EUR.

2. Growth over six months

Our assessment asks for revenue in the same month six months earlier. Andreessen Horowitz wrote in 2015 that investors often preferred the compounded monthly growth rate to a simple average: (latest month / first month) to the power of 1 / months, minus 1. From 47,400 EUR to 60,000 EUR over six months is about 4.0% a month, and 26.6% in all.

3. The share of revenue that recurs

How much of last month's revenue comes in again next month without a new sale, divided by last month's revenue. The same 2015 guide said investors valued companies more highly when the majority of revenue came from the product, because services revenue was non-recurring, had much lower margins and was less scalable. In the example, 51,000 of 60,000 EUR recurs: 85%.

4. Customers who stay, and what they pay

Of the customers who were paying six months ago, the share still paying today, and whether those same customers now pay more, the same or less. Stripe's net revenue retention starts from existing customers' recurring revenue, subtracts churn and downgrades, adds upgrades and leaves new customers out. Its guide warns that a high figure can hide lost customers, so read it next to the head count: in the example, 40 of 46 customers still pay, or 87%.

5. Your largest customer's share

The share of the last three months' revenue that came from your single largest customer. Stripe's guide to SaaS metrics calls a high concentration risky, because losing one customer would cut revenue hard. In the example the company billed 173,166 EUR over three months, and its largest customer paid 13,000 EUR of it: 7.5%.

6. Paying customers and average deal size

How many customers paid you in the last three months, and your average contract value. The 2015 guide's first question about annual contract value was its size: a few hundred dollars a month from each customer, or large deals. In the example, 48 customers at an average of 15,000 EUR a year each make 60,000 EUR a month.

7. What a customer costs to win, and how long it takes

What it cost to win one paying customer through your main channel, and your average sales cycle in months. Stripe's 2024 guide to CAC payback divides CAC by the monthly profit per customer to get the payback period. The 2015 guide said investors weighed paid CAC above a blended figure. In the example, a CAC of 4,500 EUR, 1,250 EUR of monthly revenue per customer and 250 EUR a month to serve them gives a payback of 4.5 months, on a sales cycle of three months.

8. Net burn

Total spending last month minus revenue last month. J.P. Morgan's guide defines the burn rate that way, and the 2015 guide said investors focused on net burn and on how fast revenue and costs were growing. In the example: 130,000 minus 60,000 is 70,000 EUR a month.

9. Runway, and the date you cut costs

Cash in the bank divided by net burn, with no fundraising counted in, as J.P. Morgan's guide puts it. In the example, 600,000 divided by 70,000 is about 8.6 months. Our assessment also asks for the month you will cut costs if the money you expect has not arrived. The same 2024 guide says investors look more cautiously at startups with less than six months of runway, so that date belongs next to the runway figure.

10. The round

How much you are raising, in what structure, the month you expect to close, the milestones it pays for over the next six quarters, and how much equity and debt you have raised so far. In the example: 1,500,000 EUR in a priced round, after 900,000 EUR of equity raised to date and no debt. Hold them next to the nine above, to check that the money buys the result.

How to use this page with your co-founders

Answer alone first, then compare. We recommend that every co-founder does this kind of honest check separately, before anyone sees the others' answers.

1. Each co-founder, alone, writes down all ten numbers for the last full month, with the formula they used.
2. Put the sheets side by side. Where two of you wrote different figures for the same line, find out whether you used different definitions, different months or different spreadsheets, and agree on one formula.
3. For each number, write one sentence on what it says about the company. Then read the ten sentences in order. If the story breaks at one of them, that is the number to work on before an investor reads it.
4. For any number you cannot answer yet, write down what you know, how you would estimate the rest and when you will have the real figure.
5. Do it again every quarter, on a date in the calendar that nobody moves.

Our startup assessment is built the same way. Each co-founder answers separately, and the team report shows the team average, the alignment per dimension and your answers side by side.

Work through it with us

TBHM Growth is a four-week online programme for founders whose company has revenue, paying customers and six to eight months of runway. Pre-revenue companies are not accepted. Week three is on the fundamentals of fundraising, and its session on the numbers works through acquisition cost, lifetime value, payback, gross margin, burn and runway, then the ten numbers an investor reads first. Another session that week covers how fundraising actually works, what dilution costs you and when not raising is the better answer.

The programme has eight live sessions on Zoom, four private 30-minute mentor calls, ten startups per cohort and a one-page written verdict within a week of the end. It costs 1,200 EUR per company, VAT included, co-founders join at no extra cost, and it takes no equity. See if TBHM Growth fits your company.