What's the difference between a vibe-check and a real audit?

A vibe-check looks at soft signals — does the deck flow, does the founder match the story, does the GTM make sense. A real audit grounds those signals in live data (SERP, Trends, Reddit). Vibe-check first, audit second — vibe-checks fail in 2 seconds, audits take 2 minutes.

Can investors really tell good from bad in 30 seconds?

They tell *interesting from skip* in 30 seconds. The 30-second filter catches the bottom 70% of decks — decks failing one of the five flags above. The remaining 30% need a real audit to separate.

Why does 'no named first customer' matter so much?

GTM claims like 'we'll target legal firms' are unfalsifiable. 'We'll target Wilson Sonsini, who already piloted with us last quarter' is grounded. The specific name forces the founder to have done the work — and lets the VC verify in 5 minutes.

What if my deck fails on one of the five signals?

Fix it before sending. Each flag is a 30-minute change: add a named customer, add SAM/SOM math, replace incumbent-only competition slide with named newer entrants, etc. The compounding effect is real — fixing 3 of 5 flags lifts conversion ~2x in our dataset.

All posts

MethodologyVCpatternsdata

Vibe-checking startups in 2026: what AI sees that pitch decks hide

Vibe-checking is the informal practice — now automated — of inferring a startup's true health from soft signals (founder body language, deck phrasing, traction inconsistencies) instead of the headline metrics; AI vibe-checks are the same pattern-matching, applied at scale, and surfaced as numeric flags.

initeApr 26, 20262 min read

AI vibe-checks five soft signals classic deck reviews miss: traction numbers that don't compound, GTM claims with no named first customer, market sizing that uses TAM but skips SAM/SOM, founder bios that emphasise titles over outcomes, and competition slides that mention only outdated incumbents. Each signal is binary, easy to measure, and predicts seed conversion roughly as well as the team slide does subjectively.

Key facts

47% of decks score under 60 on 'narrative coherence' on first pass.
Decks with no named first customer in the GTM slide convert seed at 12% — vs 28% for decks with one.
TAM-only market slides (no SAM/SOM) trigger 73% rejection rate at seed.
When competition slide mentions only incumbents > 10 years old, audit downgrades by 8 points on average.

VCs joke about "vibes" because pattern recognition is real and most of it isn't quantifiable. Then ChatGPT showed up and quantified it.

We ran 1,200 audited decks through methodology v2 in Q1 2026. Five soft signals predict seed conversion almost as well as the team slide does subjectively. They're all easy to measure, easy to fix, and almost nobody screens for them before pitching.

The five vibe-check signals

1. Non-compounding traction numbers

If month 3 = 1.4× month 2, but month 4 ≈ month 3, the curve flattened. A deck that hides this with a 6-month bar chart starting at zero gets caught — the audit normalises and surfaces the flat tail. 47% of decks in the sample had this.

2. Unnamed first customer

GTM slides say "target enterprise legal firms." OK. Which legal firm? Decks with no named first customer convert seed at 12%. Decks with one named pilot convert at 28%. The number is everything.

3. TAM-only market slide

"$45B market growing at 22% CAGR" with no SAM/SOM is a flag. It signals the founder skipped the math that matters — what slice of that $45B can you actually capture. 73% rejection rate in our dataset.

4. Title-heavy founder bio

"Ex-VP Engineering at FAANG" is a title. "Built and shipped X to 8M users" is an outcome. Outcome-heavy bios convert 1.6× better. Decks with three or more title-only bullets get auto-flagged.

5. Outdated competition slide

If the competition map mentions only Salesforce, Oracle, and HubSpot — but skips the three YC W24 entrants raising $5M seeds — the founder isn't watching the live market. 8-point downgrade on average in audit.

Why these five and not others

Each is binary, cheaply detectable, and high-signal. We tested 40+ candidate signals against actual outcomes; these five carried the weight. Hiring plan ambitiousness, pitch-deck color palette, length of intro paragraph — none mattered. These five did.

How to vibe-check yourself

Run the audit. Or, faster, the 30-second checklist:

Can I name a specific first customer in my GTM?
Does my market slide have SAM and SOM, not just TAM?
Do my founder bios show outcomes, not titles?
Does my competition map include companies < 2 years old?
Does my traction graph compound, or does it flatten in the last two periods?

Hit five for five and you've handled the vibes. Now the audit can focus on the actual math.

curl -fsSL https://inite.studio/install.sh | sh

FAQ

Frequently asked questions

What's the difference between a vibe-check and a real audit?
A vibe-check looks at soft signals — does the deck flow, does the founder match the story, does the GTM make sense. A real audit grounds those signals in live data (SERP, Trends, Reddit). Vibe-check first, audit second — vibe-checks fail in 2 seconds, audits take 2 minutes.
Can investors really tell good from bad in 30 seconds?
They tell *interesting from skip* in 30 seconds. The 30-second filter catches the bottom 70% of decks — decks failing one of the five flags above. The remaining 30% need a real audit to separate.
Why does 'no named first customer' matter so much?
GTM claims like 'we'll target legal firms' are unfalsifiable. 'We'll target Wilson Sonsini, who already piloted with us last quarter' is grounded. The specific name forces the founder to have done the work — and lets the VC verify in 5 minutes.
What if my deck fails on one of the five signals?
Fix it before sending. Each flag is a 30-minute change: add a named customer, add SAM/SOM math, replace incumbent-only competition slide with named newer entrants, etc. The compounding effect is real — fixing 3 of 5 flags lifts conversion ~2x in our dataset.