I Built a Self-Evolving AI Sidekick for Search — Offering Setup as a Service

If you've spent any time in search, you know the information problem. You're one person, or maybe two, trying to track hundreds of target businesses, research a dozen industries, keep investors updated, run DD on three simultaneous prospects, and somehow also do outreach. I've been building something that I think is a genuinely useful tool for this: a self-evolving autonomous agent that runs on a schedule, accumulates knowledge over time, and is guided by whatever vision and goals you set for it. The basic idea, you give it two things: - A VISION — a 2–4 paragraph description of what you're trying to accomplish. For a searcher: your target industry, deal size, geography, what a great acquisition looks like, what you're ultimately trying to build. - A set of GOALS — the 2–4 directions you want it working on right now. These might be "research the HVAC services market in the Southeast," "compile a list of business brokers active in sub-$5M transactions," or "draft a monthly investor update format." From there, the agent runs sessions on a schedule, hourly, daily, whatever you want, and self-directs. It reads its own memory from previous sessions, updates its goals when directions are complete or stale, and always produces committed, reviewable output. It's not a chatbot. You don't interact with it session-to-session. It just works, like a junior associate who shows up every morning, reads everything in the file, and makes progress. What I'm offering: I'll set this up for you, configured for your specific search, for a flat fee. Takes about an hour on my end. You get a GitHub repo, a cron job, and an agent that knows your criteria. Happy to answer questions about what it can and can't do. More specific posts to follow on deal sourcing, DD research, and investor comms use cases. Technology: Claude Code CLI + bash. Runs on any Linux/Mac server or even a cheap VPS. No subscription to the agent itself, just your Claude API usage.