Hi everyone,

For Sprint 7, I’ve been working on how I use AI inside our actual VC workflow at NextWorld Capital Fund I, and I’d really value your feedback on the setup and where you think it could be stronger.

I built a dedicated ChatGPT Project for NextWorld Signal Fund I as an AI-assisted operating workspace for live venture work. It is connected to the Fund’s Internal Data Room and uses custom instructions covering the investment thesis, evidence standards, governance, portfolio construction, LP fundraising discipline and AI boundaries.

I also use Venture Scouting, a dedicated ChatGPT Project for reviewing founder-submitted startup information. It turns founder inputs into a structured scouting assessment covering fund fit, founders, product, traction, market, business model, risks, missing evidence, investment readiness and next steps.

A strong example was a live underwriting review of FITINION, a Pre-Seed opportunity in our pipeline. The AI did more than summarize the company. It separated Claim / Documented / Verified evidence, kept the company correctly at the Diligence stage rather than treating it as an approved investment, identified missing product and traction evidence, flagged a potential conflict around NextWorld’s pre-existing 5% position, tested ownership against the Fund’s 10–12% planning target, and connected the opportunity to the Fund’s power-law return framework.

The part I found most useful is that AI can combine company-level information with fund strategy, governance, evidence standards and portfolio economics in one review. I see AI as especially useful for evidence organization, screening, diligence-gap identification, research synthesis, portfolio modeling and decision documentation. I see it as least useful for final investment judgment, founder trust assessment, legal/compliance decisions and anything that requires human accountability.

My biggest current gap is automation. Scout is still human-operated, so founder information is entered manually and downstream actions such as CRM updates are not yet automatic.

I’d really appreciate feedback on three things:

Thanks, direct feedback is very welcome.