Nobody is born knowing how a round works.

An investor hands you cash and receives a permanent slice of whatever the company becomes. There is no repayment schedule and no interest to negotiate away. The only way that slice pays them back is if the company is worth a great deal more later, and someone buys it or the company lists.

That is why questions that feel hostile — how big can this get, why can't someone copy it, what happens when a large AI company notices — are not doubts about you. They are the only questions that decide whether the slice is worth anything.

• Idea / pre-product: the money buys a first working version and the first proof that someone wants it. Investors are judging the team, the insight and whether the problem is real.
• Pre-seed: the money buys evidence — a product in real hands, early usage, the first signs of pull. Investors are judging whether the insight survived contact with users.
• Seed: the money buys a repeatable way to acquire customers. Investors are judging whether growth is a machine or a series of favours.

Claiming a later stage than your evidence supports does not raise your valuation. It raises the bar you are measured against, and you fail it with your own material.

• Is the prize big enough? Not the size of the industry — the size of the part you can actually sell to.
• Why you? What do you know, own or have built that a competent stranger does not.
• Why now? What changed in technology, regulation or behaviour that makes this possible this year and not five years ago.
• What does the money buy? Which specific milestone this round reaches, and why that milestone makes the next round raisable.

An early-stage valuation is not calculated from your revenue. It is set by what the next investor will plausibly pay. If you price this round so high that the next round cannot clear it, you have not won anything — you have set up a down round, which damages you far more than a modest price today.

• SAFE: money now, shares later. Fast and cheap, no interest, no maturity date. The price is settled at the next priced round.
• Convertible note: the same idea as a loan — it carries interest and a date by which it must convert or be repaid. That date can become a crisis if the next round slips.
• Priced round (equity): shares are issued today at an agreed valuation. Slower and more expensive in legal work, but everyone knows exactly what they own.

Two terms decide what a SAFE or note really costs you. The discount lets the early money convert at a lower price than the next round. The valuation cap sets a ceiling on the price it converts at, no matter how well the next round goes. Stack several capped SAFEs and you can be far more diluted at the priced round than you expected. Model it before you sign, not after.

Every one of these words is in the glossary if you want the one-line version.

• Write down what the next round's investors will need to see before they price you higher.
• Cost out the work that gets you there, honestly, including the hires.
• Add the months it will take to raise the next round — that search is itself a cost.
• That total is your raise. If it needs more than roughly eighteen months of runway, the milestone is probably too ambitious for one round.

A round with no named milestone reads as "we will figure it out". That is the single easiest thing for an investor to decline.

• Deck: the problem, why now, what you built, who uses it, the plan for the money, the team.
• Financials: what you actually earn and spend, how many months of cash remain, and the assumptions behind any forecast.
• Technical material: what is genuinely built, what is bought or wrapped, and what would be hard for a competitor to reproduce.
• Cap table and existing terms: who owns what already, and any caps or discounts still outstanding.

On this platform, anything the material never addresses is marked "not covered" and left out of the score rather than counted against you. Silence is not punished — but it is also not credited.

• Could a general model do this next year? If your product is a thin layer over something a large model already almost does, say how you stay ahead of it.
• How long would a competent team need to copy you? Measured in weeks or in years — and be specific about what makes the difference.
• What are you accumulating? Data, distribution, workflow lock-in, regulatory clearance. Something that gets harder to catch the longer you run.
• How much of the raise goes into building software? If most of the money is engineering that a small team plus AI tooling could now do far cheaper, that is a question about the plan, not about you.

• A market size taken from an industry report with no line connecting it to what you sell.
• A list of features where a defensibility argument should be.
• A use of funds that does not add up to the milestone being promised.
• A team with no visible reason to be the ones who win this particular market.
• Forecasts with no stated assumptions, which makes the whole financial section unusable.
• No answer at all to the AI question — which the reader then answers for you.

Numbers above are as publicly reported and rounded; treat them as orientation, not benchmarks. What matters is the shape of the read: the same handful of questions, answered by evidence that existed at the time. That is exactly what your report does to your material — which is why uploading the financials and the technical files alongside the deck changes the read so often.

• Code is no longer a moat on its own — AI writes it too cheaply. What compounds (supply, data, relationships, regulatory positions, trust) is what survives the read.
• Polished narratives are now free to produce, so evidence discipline matters more, not less: documents and observed behaviour outrank a well-written deck.
• Every deal now carries an explicit displacement question — would an AI tool or an AI-native incumbent absorb this within a year or two? Your report scores that question by name, for every project, instead of leaving it to chance.

Each judgement in your report carries an evidence label: taken from a document, partially supported, a founder claim, or not covered. A low score attached to "founder claim" usually means the belief is fine and the proof is missing — that is a document to add, not a business to abandon.

The report also gives you the gaps in your material, the milestone the next round will expect, and the conditions that would prove the report itself wrong. Work the list, upload again, and the read changes.

Term you do not recognise? The glossary covers every one used in a report.

• Send the same short operating update every month. Comparable numbers build more trust than a newly polished deck.
• Share bad news when it becomes visible, not when you have made it presentable. Bring what you know, what you do not know and when you will know more.
• Make requests executable: name the person or profile, why the fit matters, the action you want and the date it is needed.
• Take a hard question seriously without surrendering the operating decision. Ask what evidence would change the investor's view, then decide whether the test is worth running.
• Use a Portfolio Health action list to improve the company, not to optimise the score. Pick one to three priorities with owners, dates and a signal that proves the concern wrong.

The full shared workflow — meeting cadence, roles, introductions, disagreements, AI-era reviews and copyable update templates — is in the Founder × Investor partnership guide.