Every founder preps for VC meetings the same way: run it by an advisor, pitch it to a founder friend, rehearse it in front of a mirror. All three have something in common that nobody points out — none of them want you to fail. An advisor softens the follow-up question. A friend moves on when you look uncomfortable. A pitch coach has heard the same slide so many times they've stopped pushing on it.

A partner at a fund evaluating your round has no such constraint. They've sat through hundreds of pitches in your category. They already know what a weak answer to their next question looks like, and they're watching to see if yours is one. They won't move on when you hesitate. They'll push.

That gap — between how you practice and how the real meeting actually goes — is where deals die before the founder ever finds out why.

Fundraising has always involved rejection. What's changed is how long you're now expected to survive on your own preparation before anyone writes a check.

The median time between a seed round and a Series A has nearly doubled in three years. Investors are writing fewer checks, taking longer to decide, and asking for more evidence before they commit — which means every meeting you get carries more weight than it did in 2021, and you get fewer of them to make your case.

And when you do get the meeting, the deck itself has almost no room for error.

Investors spend under four minutes on a first look at a seed deck, and more than four in ten don't make it to the last slide. There is no version of "I'll explain that better in the room" that survives those numbers. The questions that would have killed the deal need to already have good answers, because there usually isn't a second pass.

The problem isn't effort. Founders who lose deals aren't under-prepared in the sense of not trying — they've rehearsed, they've gotten feedback, they've tightened the deck. The problem is what they've rehearsed against.

An advisor, a friend, a coach — all of them are, on some level, on your side. That's exactly what makes them unable to replicate the one thing a real investor brings to the table: a strong, informed prior that your company is going to fail the same way the last twenty in your category did, and no obligation to be gentle while they test whether that's true.

This is where a growing number of founders have started using AI differently than most people default to using it. The instinct with a tool like Claude is almost always to ask it to help — tighten this paragraph, summarize this doc, be encouraging about the idea. Pointed the other way, given an explicit adversarial role instead of a helpful one, it does something a mirror or a friend can't: it holds a skeptical position across an entire conversation, with no social pressure to soften and no reason to move on until an answer actually holds up. Paste a deck into a fresh chat with no framing and you'll get exactly the encouraging feedback you'd expect. Set the frame first, and what follows starts to resemble the meeting you're actually walking into.

That's the system below: a setup prompt that fixes the adversarial role, a sequence of challenges built around the categories investors actually evaluate, and a closing prompt that turns the whole session into a prep sheet you can read before every real meeting.

Run these in order, in a single chat — don't start a new conversation partway through. The value compounds: every hedge, every vague answer, every assumption you couldn't quite defend becomes part of the context the later prompts draw on. Be honest in what you paste in. It's tempting to hand over the best version of your metrics; the more accurate your input, the more useful the session actually is.

Run this first, every time, before any of the challenge prompts.