PITCH DECK REVIEW · IN TELEGRAM

Know why investors pass. In 10 minutes.

Send your deck to the bot. Get back a verdict, a 0–100 score and slide-by-slide comments showing exactly where investor trust breaks, and how to fix it.

Review my deck ($5)

Delivered in Telegram as a full report, usually in about 10 minutes. Useless? Full refund, no questions.

Sample: ElevenLabs pre-seed deck · 62/100 · needs work

We ran it on a famous deck. Read every word.

This is the deck ElevenLabs used to raise $2M before becoming a multi-billion-dollar company (the public copy has traction and timeline redacted). Our review still found exactly where a cold investor read breaks. Yours gets the same treatment.

PDFelevenlabs-preseed.pdf

14 slides · 2.4 MB

Got it. Reviewing 14 slides now, this usually takes about 10 minutes.

HTMLPitch deck review — Eleven

Verdict: needs_work · 62/100Full report attached: 9 category scores, 42 slide-by-slide comments.

Inside the report: verdict, takeaways, category scores

Zoom into one slide. The market slide, and what the review pinned to it:

1$2B / $4.6B / $24B anchors, no math

Methodology is not visible: the reader cannot evaluate how these numbers were calculated or whether they map to the obtainable wedge.

2Category boundary feels loose

Professional creators, game localization, movie dubbing, translation and interpreting are not yet tied into one credible beachhead.

3The proof trail is the problem

The deck can start a conversation, but it needs stronger visible proof before it can carry a cold investor read.

Read the full sample review → See the original deck →

What happens after you pay

The whole exchange lives in one Telegram chat. You are never waiting in the dark.

STEP 1 · 30 SECONDS

Pay $5

Telegram Stars, inside the chat. No signup, no email, no card form.

STEP 2 · 10 SECONDS

Attach your deck

PDF, up to 15 slides. Longer deck? The bot asks which pages to review.

STEP 3 · ~10 MINUTES

The pipeline runs

Every slide is read, scored against the rubric, and the report passes an automated quality gate.

STEP 4 · DONE

The report arrives

One message: verdict, 0–100 score, category scores, slide-by-slide comments. Like the sample above.

Comments this specific, on every slide

Six of the 42 comments from the ElevenLabs sample, word for word.

COVER SLIDE

Investor context is missing: round, ask, audience, and current company stage are not visible on the opening slide.

INTRO SLIDETRUST BREAK

The ~$100/min and >2 weeks numbers are load-bearing, but no visible citation or methodology supports them.

SOLUTION SLIDE

“Human quality” is the core promise, but the deck does not yet prove it with examples or validation.

TRACTION SLIDETRUST BREAK

After market sizing, the reader needs proof; the redacted block interrupts the argument at the worst point.

TEAM SLIDEWORKING

Google, Palantir, Cambridge/Oxford/Imperial, NeurIPS and open-source signals are relevant to the technical challenge.

CLOSE SLIDE

The close does not restate round, raise amount, use of funds, or desired next conversation.

What $5 actually buys

Not one prompt. A pipeline that reads your deck the way an investor does, then checks its own work.

1IngestEvery slide rendered as an image and OCR’d; nothing is skimmed.

2Slide mapWhat each slide claims, and what it must prove to a cold reader.

3Score35-point checklist across 6 lenses: narrative, market, proof, revenue, GTM, design.

4Cross-checkClaims tested against investor-pattern rules from real deck teardowns.

5Quality gateThe report itself is machine-checked before it reaches you. Fails? It reruns.

~1.5Mtokens processed per review

210checklist evaluations per deck

~25analysis steps per run

100%of reports pass the quality gate before delivery

Then hand the fixes to your AI

The report is written to be actionable by the tool you build your deck with. Paste it into ChatGPT, Claude, Gamma or Pitch together with your deck, and let it apply the fixes slide by slide.

Here is a professional review of my pitch deck. Go through the slide-by-slide comments and rewrite each flagged slide: fix what the review calls a trust break first, keep my facts, don't invent numbers.

[attach: your deck + the review]

Copy that prompt as-is. The review's structure (verdict, categories, per-slide comments) is what makes it work as AI input.

Why this isn’t “just ChatGPT”

The model is table stakes. The product is the rubric: 50 investor deck-teardown sessions transcribed, 180+ timestamped rulings coded into scoring rules, checklists from YC and 500 Startups, 13,000+ lines of review methodology, and an automated quality gate on every report.

A chatbot gives you opinions and a 78. This gives you the exact slide where investor trust breaks, scored on a harsh curve: the famous deck above got a 62.

The price is not the punchline

It’s the same class of review, without the human calendar attached.

HUMAN ROAST SERVICES$100–350

Days of turnaround. Quality depends on who picks up your deck.

AI COMPETITORS$29–49

Mostly a score and generic advice. Samples rarely shown, for a reason.

THIS REVIEW$5

Full report in ~10 minutes. Useless? Full refund, no questions asked.

Review my deck ($5)

Questions founders ask

Is this just ChatGPT with a system prompt?

No. It is a multi-step pipeline: slide rendering and OCR, a slide map, a 35-point rubric across 6 lenses calibrated on transcribed investor teardowns, claim cross-checks, and an automated quality gate that blocks weak reports. About 1.5M tokens of work per review.

Is my deck confidential?

Yes. Your deck is never shared, published, or used as a sample. Public samples like the ElevenLabs review are made only from decks that are already public.

What do I get, exactly?

An HTML report in Telegram: verdict (ready / nearly ready / needs work / not ready), 0–100 score, 9 category scores, and slide-by-slide comments. Read the sample above; yours has the same depth.

What formats and sizes?

PDF, up to 15 slides reviewed per run. For longer decks the bot asks which page range to focus on.

What if the review is useless?

Full refund, no questions. So far the refund count is zero.