Your deck is read by a machine before a partner sees it
85% of dealmakers now run AI on daily work and 82% use it for sourcing. Your deck gets summarised and scored before a human opens it. Write for that pass.
An analyst at a decent fund sees around three thousand decks a year. The fund invests in roughly nine of them.
Do the arithmetic on the reading time. Two to three minutes a deck, and that is before anything gets read properly.
That job stopped being done by a person some time ago.
What actually happens to your file
It gets opened by something that extracts fields. Team, product, traction, market, ask.
Those fields get scored against the fund’s own rubric. The output is three sentences and a routing decision.
A partner reads the three sentences. Sometimes they open the file. Often they do not.
None of this is hostile. It is what you would build too, if nine of every three thousand mattered.
The mistake is writing for the reader who comes second.
The extraction pass is unforgiving in one specific way
It cannot infer. If a number is on a chart with no label, it does not exist.
If your revenue lives in a slide footnote, it comes back as UNKNOWN. If your ask is only in the email body, the deck scores as having no ask.
A human skims past that. They fill the gap from context, or from liking you.
The extraction pass fills nothing. Whatever is not stated is missing, and missing fields lower the score.
Same deck, two readers
Written for a person
Context does the work
- A growth chart with no axis labels
- ARR implied across three slides
- The ask sits in the covering email
- Customer logos, no contract status
Written for the pass
Every claim is a field
- The number written next to the chart
- ARR stated once, in figures, with the date
- The ask on the deck, with use of funds
- Logos labelled: paid, pilot or letter of intent
This is not about writing for robots
I am not telling you to strip the story out. The story is what makes a partner open the file.
I am telling you the story now has to survive being reduced to fields first.
Both readers exist. One of them decides whether the other one shows up.
The fix is boring. Put every claim in a place a parser can find it, then let the narrative sit on top.
Fields that go missing
- The number, in figures Written as text next to the chart. Not implied by the shape of a line.
- The date on the number A metric with no month attached gets read as stale or gets skipped.
- The status of a logo Paid, pilot, trial or letter of intent. A wall of logos with no labels scores as nothing.
- The ask and the use Amount, stage, runway it buys. If it is only in the email, the deck has no ask.
Run the pass on yourself first
You can do exactly what they do. Same rubric, same refusal to infer.
Export your deck to text or images, paste it in, and read what comes back.
You are a deal screening agent at an early stage venture fund. You process
around 3,000 decks a year and pass on most of them within two minutes.
Here is the deck:
[PASTE THE DECK TEXT, OR ATTACH THE FILE]
Extract these fields. Use ONLY what the deck states. Never infer a number
from a chart shape, never fill a gap from what you know about the market,
and never use the company website. Write MISSING for anything absent.
COMPANY:
ONE_LINE:
STAGE:
ASK_AMOUNT:
USE_OF_FUNDS:
REVENUE_OR_TRACTION: (figure and the date it refers to)
CUSTOMERS: (named, with status: paid, pilot, or letter of intent)
MARKET_SIZE_CLAIM: (and whether the deck sources it)
TEAM_RELEVANCE: (why this team, in the deck's own words)
COMPETITION:
RISK_THE_DECK_ADMITS:
Then write:
1. A three sentence summary, the kind a partner reads instead of the deck.
2. Every field you marked MISSING, as a list.
3. One line: on this evidence alone, would you route this to a partner or
to the pass pile? Answer PASS if three or more fields are MISSING.
Read the MISSING list, not the summary
The summary will flatter you. It is written from your own words.
The MISSING list is the finding. Those are the fields a real screen returns empty on your deck.
Fix those slides, run it again, and watch the list shrink. When it hits one or two, your deck survives the pass.
Then go back to worrying about the story, which is the part that was always the hard bit.
The screening prompt is on the resources page. Free, no email required.