Venture Capital
$17M Deployed Across 326 Startups Through Automated Deal Flow
A deal pipeline that runs itself, and an evaluation engine that gets sharper with every decision.
The challenge
Ignite Ventures had capital to deploy and a deal pipeline that couldn’t keep up with it. Everything was manual. Incoming investors were qualified by hand, one conversation at a time. Startups were reviewed individually, each evaluation starting from a blank page. Discovery calls were scheduled through back-and-forth email.
Critically, the process had no feedback loop: the two hundredth evaluation was performed with exactly the same judgment infrastructure as the first, because nothing the firm learned ever made it back into how the next deal was assessed.
At the volume Ignite was targeting, this became the binding constraint on capital deployment. Not the capital, not the deal flow: the processing capacity. Good startups sat in a queue. The bottleneck wasn’t strategy, it was throughput.
What we built
A fully automated deal flow system that runs the investment pipeline end to end. No human touches a deal until it reaches final review.
Automated investor qualification
Incoming investors are qualified automatically against the firm’s criteria, routing serious prospects forward and filtering the rest without consuming partner time.
Configurable startup scoring
Every incoming startup is scored against a structured, configurable rubric: the same criteria applied consistently to deal one and deal three hundred. The rubric re-weights as the thesis evolves.
RAG-powered evaluation that learns
A retrieval-augmented layer draws on the firm’s own history of decisions. Every decision , funded, passed, and why , becomes context for the next evaluation. Scoring accuracy improves the more the platform is used.
Automated discovery scheduling
Startups that clear the scoring threshold have discovery calls scheduled automatically. The gap between “this looks interesting” and “we’re talking” collapses to a booked slot.
How it works
Intake: investor and startup submissions enter through a structured pipeline.
Qualify: investors are assessed against firm criteria and routed.
Score: startups are evaluated against the configurable rubric.
Retrieve: the RAG layer pulls relevant precedent from prior decisions to inform the score.
Schedule: qualifying startups are automatically booked into discovery.
Review: a human sees the deal for the first time, with a scored, contextualised brief attached.
Learn: the outcome is written back into the knowledge base, improving the next evaluation.
Results
$17M deployed across 326 startups
Deal evaluation decoupled from partner headcount
Consistent scoring criteria applied across the entire pipeline
An evaluation engine that gets measurably better with use
Partner time redirected from screening to closing
See it in action
Watch the deal flow system process a live submission: automated scoring against the rubric, RAG retrieval pulling precedent, and discovery scheduling firing automatically.
Technology
FAQ
How does the RAG layer actually improve scoring over time?⌄
Each completed decision , including the reasoning behind a pass , becomes part of the knowledge base. Future evaluations retrieve relevant prior decisions as context, so the model scores against the firm’s demonstrated judgment rather than a generic template.
Can the scoring rubric be changed after launch?⌄
Yes. The rubric is configuration, not code. Criteria and weightings adjust as the thesis shifts, without touching the underlying system.
Does this replace investment judgment?⌄
No, it replaces screening. The system handles qualification, scoring and scheduling so partners spend time on final review and closing, where judgment creates value.
What volume can it handle?⌄
The pipeline isn’t headcount-bound, so throughput scales with deal flow rather than team size. 326 startups were scored through this system.
Is your deal flow bottlenecked by process rather than capital?
If good deals are sitting in a queue behind manual screening, there’s a pipeline that runs itself.