Turning Manual Lead Research into a Working Tool
A working internal MVP. No usage data yet.
Problem
- Lead research was manual. Finding companies, judging fit, and preparing first-outreach context took repeated effort.
- Optifeed relied on inbound, founder-led sales, or manual research to find ICP-fit e-commerce brands.
- Marketing needed a repeatable loop: create a project, run a search, get explainable candidates, and review them before any outreach.
The smallest useful loop
Not a GTM automation platform. Just this:
- Create. A project or campaign.
- Search. Run a lead-research search.
- Generate. Claude returns explainable company candidates.
- Review. Results are saved to the project page for marketing to review before any next step.
What I did
- Found the marketing problem, formed the hypothesis, and mapped the UX flow in Miro.
- Scoped the direction with the CTO before committing to the MVP shape.
- Built the MVP in Rails with the Anthropic API: project and search structure, saved results, and project-page views.
- Used Claude Code and Codex for UI work and build support. How I use AI →
Four decisions
One project per campaign.
A repeatable structure, not one-off searches.
Explain every candidate.
Not a raw company list.
Keep the results.
Saved, not a temporary AI response.
People decide.
Marketing reviews before any automated action.
Where it stands
When my engagement ended, OptiGTM was a working internal MVP. I have no usage or adoption data.