Affinity Logs Deal Flow. It Doesn't Read the Pitch Decks.
Key takeaways
- Affinity maps a fund's network and logs deal-flow stage — it doesn't open an incoming pitch deck and tell an analyst if it fits the thesis.
- Analysts spend a meaningful share of their week on first-pass screening: opening a deck, extracting the details, comparing against criteria they already know by heart.
- An n8n + Claude agent extracts the thesis-relevant details from every incoming deck and screens it against the fund's actual criteria before an analyst opens a slide.
- Rejected decks get a documented reason, not silence — a partner can override the screen on any deal, any time.
- Funds running this get their analysts spending time on deals that already cleared a real first screen, instead of being the screen themselves.
Affinity maps a fund's network.
Tracks relationship warmth. Logs deal-flow stage.
What it doesn't do — what no relationship-intelligence platform is built to do — is open an incoming pitch deck and say whether it fits the thesis.
That first read is still a human, opening a deck cold, every single time.
Where the Real Analyst Hours Go
A fund receiving real inbound volume has analysts spending a meaningful share of their week on first-pass screening: opening a deck, extracting the market, the traction numbers, the team background, the ask, and comparing it against criteria the fund already knows by heart. Most of that volume doesn't fit the thesis and never will. The problem isn't that screening is hard — it's that it's repetitive, and repetitive judgment against a known, stable set of criteria burns analyst hours without requiring an analyst's actual expertise, until the genuinely promising decks surface.
Here's Exactly What the Agent Does Before an Analyst Opens a Slide
1. Ingest. Incoming decks are pulled the moment they arrive, whatever the source.
2. Extract. Claude pulls the thesis-relevant details — market, stage, traction, team, ask.
3. Screen. Every deck is compared against the fund's actual investment criteria, not a generic model of "what a good deal looks like."
4. Recommend. A structured summary and a clear recommendation are produced for every single deck.
5. Route. Deals that clear the screen go straight to an analyst's full attention. Deals that don't get a documented reason, not silence.
What This Actually Replaces
Not the investment decision — that stays with the partners and analysts who make it. What it replaces is the first-pass reading labor: the hours spent opening decks that were never going to clear the bar, done identically every time by whoever's turn it is in the inbox rotation. Analysts spend their attention on deals that already passed a real first screen, instead of being the screen themselves for every inbound deck.
Where Deal-Screening Automation Falls Apart
Mistake one: screening against criteria too vague to score. "Good team" isn't a screenable criterion. Funds that don't translate their thesis into specific, checkable signals get a screen that either rejects everything or nothing.
Mistake two: no feedback loop from partner overrides. If a partner overrides the screen and nobody records why, the screen never improves — it just keeps making the same near-miss over and over.
Mistake three: one screen across very different deal types. An early-stage SaaS thesis and a growth-equity thesis need different criteria. A single screen tuned for one systematically misjudges the other.
How Funds Actually Roll This Out
Weeks 1-2: Translate the fund's actual thesis into specific, checkable criteria — this step alone often surfaces disagreement among partners about what the thesis actually means in practice.
Weeks 3-4: Run the screen in parallel with the existing manual process, comparing outcomes before turning off manual screening entirely.
Ongoing: Log every partner override with a reason, and revisit criteria quarterly as the thesis evolves.
Translating a fund's actual thesis into screenable criteria is exactly the kind of work a PE- and VC-focused automation partner should be doing before a single deck gets screened.
Frequently Asked Questions
Does this make the investment decision?
No. It screens against known criteria and produces a recommendation with reasoning — partners and analysts still make every actual investment decision.
Does this replace Affinity?
No — Affinity remains the system of record for relationships and deal-flow tracking. The workflow reads decks and can write results back into Affinity; it doesn't replace what Affinity tracks.
What happens to decks the screen rejects?
They're flagged with a specific, documented reason rather than silently discarded — so a partner can override the screen on any deal at any time.
Does this work for funds with an unusual or narrow thesis?
Yes — the criteria the agent screens against are the fund's own, configured explicitly; it isn't screening against a generic model of what a "good deal" looks like.
What if partners disagree about the actual criteria?
That disagreement usually surfaces during setup, before the screen ever runs — which firms tell us is useful on its own, independent of the automation.
Does this handle different deal types differently?
Yes — criteria are configured per deal type or stage, not as one blanket screen across an entire fund's mandate.
Get new articles when they publish
One email per post. No pitch, no spam.
