How Two Startup Funds Cut Deal Diligence to Under 4 Hours
Analysts at two early-stage funds spent 40 to 60 hours on each first-pass diligence. An AI platform now scores every deal against the fund's own framework in under 4.
40-60 hrs → <4
Analyst hours per diligence pass
4–8 weeks
Implementation Time
Project Cost
the challenge
Two early-stage investment funds, one private equity and one venture capital, were seeing deal flow grow faster than their analysts could keep up. Each first-pass diligence review took 40 to 60 analyst hours: reading the data room, researching the company and writing up a view against the fund's criteria. With small teams, every extra hour of diligence meant a deal that never got a proper look.
what they built
Portfolio Leverage built DueDrill, an AI due diligence platform that produces a structured first-pass review against each fund's own framework.
The review covers 16 categories and 214 fields. Each category is scored with a weighted rubric, and the scores roll up into an overall verdict ranging from Strong Pass to Hard Pass. The output is an investment-committee-ready memo with cited, editable fields, available as a live web view, a PDF and a shareable link.
Analysts review and refine the AI's first pass instead of building it from scratch, and the model used for each category can be switched between providers.
The build started with the rubric, written by hand before any code: 16 categories, 214 fields and the weights that decide how each one counts. Next came a single AI layer wrapping Claude Sonnet 4.5, GPT-5 and Gemini 2.5 Pro, with shared request handling and structured-output shims so any category can run on any model. A scoring engine rolls category scores into a weighted overall score mapped to verdicts from Strong Pass to Hard Pass. One section definition drives three outputs, the live report, the PDF and the share link, so they never drift apart. Authentication, access controls and rate limits came last, along with logging of every AI call and analyst override. The build took about six weeks. The hardest problems were operational: provider configs that had drifted apart, share links that broke on a read-only production filesystem, and auth checks that needed strict timeouts. The core lesson was that the rubric is the product and the model is plumbing.
best fit for
Small PE and VC funds, search funds and corporate development teams whose deal flow has outgrown analyst capacity, and who already have, or can write down, a clear framework for what makes a deal worth pursuing. It works best when analysts stay in the loop to review and refine the first pass.
Ai ROLE
The models research and assess a target across 16 categories and 214 fields, fill in cited fields, and score each category against the fund's rubric. The platform rolls those into a weighted verdict, and analysts review and edit the result.
impact
40-60 hrs → <4
Analyst hours per first-pass diligence, averaged across both funds.
3x
Deal-evaluation capacity with no added headcount.
16 categories, 214 fields
Scored against each fund's own weighted rubric in every review.
implementation complexity
Medium: a hand-written 214-field rubric, a multi-model AI layer, a weighted scoring engine and three output formats driven by one definition, built in about six weeks.
Yuri Kruman
Founder at Portfolio Leverage | Fractional AI Operating Partner
Portfolio Leverage Company
Founder of Portfolio Leverage, a fractional AI operating partner for PE-backed and mid-market companies. Three-time CHRO with a JD who builds the AI systems himself.
How did two startup funds cut first-pass due diligence to under 4 hours with AI?
They used an AI platform that scores each target across 16 categories and 214 fields against the fund's own weighted rubric and drafts an investment-committee memo for analysts to refine. First-pass diligence fell from 40 to 60 analyst hours per deal to under 4, averaged across both funds.
What AI models were used for the due diligence platform?
A shared multi-model layer running Claude Sonnet 4.5, GPT-5 and Gemini 2.5 Pro, built on Next.js, Vercel and Supabase.
What results did the funds see?
Analyst time per first-pass diligence dropped from 40 to 60 hours to under 4, and deal-evaluation capacity tripled without adding headcount. The figures are averages across the two funds.
How long did the AI due diligence platform take to build?
About six weeks, starting with a hand-written rubric before any code.
Who is an AI due diligence platform best for?
Small PE and VC funds, search funds and corporate development teams whose deal flow has outgrown analyst capacity and who can define a clear framework for evaluating deals.
Have a similar challenge?
Ask whether this would work for you, or describe what you're trying to solve.