SparrowGenie

What it does

SparrowGenie is an AI sales platform. The part I owned, RFP Automation, takes an incoming RFP and generates the answers with AI, so the BDRs, AEs and Proposal Managers on a deal review answers instead of writing every one from scratch.

Why I built it

I took the RFP Automation module from 0 to 1, starting in June 2025.

It started with people, not features. 10+ interviews with BDRs, AEs and Proposal Managers. 12 end-to-end journeys mapped. Out of those came 7 coordination breakdowns that kept recurring, and those set the build order. That became 8 PRDs and 30+ scoped user stories.

On the AI side, I defined the prompt logic behind answer generation and evaluated the LLM's output against accuracy and relevance benchmarks. Every failure fed into the next version.

In beta it ran 30+ RFPs end-to-end, with 5 users across 2 teams. The published case study showed it answering 90% of RFP questions and cutting infosec and sales review time by 30%.

What I'd do differently

It worked. I made the case to shelve it anyway.

Answer quality depended on how dense the customer's knowledge base was. It took 4 to 6 weeks of regular use, feedback and training before the output was customer-ready. That pushed time-to-value and the deal cycle past what acquisition cost and runway could carry.

I took the usage and pipeline data to the CEO and the BU Head, and the module was shelved in April 2026 on that basis.

What I'd change is when I asked the question. Time-to-value was measurable from the first beta account, and I should have tested it before the build was finished, not after the beta proved the product worked. I now apply that test to everything on the SparrowCRM roadmap.