All workInternal MVP

OptiGTM · Optifeed · 2026

Turning Manual Lead Research into a Working Tool

An internal AI lead research MVP built with Rails and the Anthropic API

01 Status

Current status

Internal MVP / internal tool. Actively evolving. Usage data is not yet available. Production adoption is not claimed. Repo is private.

02 Summary

OptiGTM is an internal MVP for researching and evaluating ICP-fit e-commerce brands at Optifeed.

The problem was manual lead research: finding relevant companies, understanding fit, and preparing enough context for a first outreach took too much repeated effort.

I worked from marketing problem discovery through UX flow, CTO scoping, and Rails MVP build with Anthropic API integration.

The tool is an actively evolving internal MVP; production adoption and usage metrics are not claimed.

03 Problem

Optifeed needed a better way to discover and evaluate ICP-fit e-commerce brands instead of relying only on inbound, founder-led sales, or manual research.

The marketing/sales workflow needed a repeatable way to create a research project, run a search, receive explainable company candidates, and review the result before any outreach decision.

04 Key Constraint

MVP boundary

The goal was not to build a complete GTM automation platform. The goal was the smallest useful loop: create project, run search, return explainable company candidates, save/show results, and let marketing review.

05 My Role

  • Discovery and hypothesis
    Marketing problem discovery, hypothesis formation, and UX flow in Miro.
  • Technical scoping
    Scoped the direction with CTO context before committing to the MVP shape.
  • MVP build
    Rails MVP work at practical prototype/MVP level with Anthropic API integration understanding.
  • AI-assisted implementation
    UI/view updates, result saving/display flow, and build support with Claude Code and Codex.

06 MVP Flow

  • Create
    Create a project or campaign.
  • Search
    Run a lead research search.
  • Generate
    Claude / Anthropic returns explainable company candidates.
  • Review
    Save the result, show it on the project page, and let marketing review before deciding what to do next.
OptiGTM project page showing a saved lead research search result
Project resultShows the project-based review loop without exposing internal data.

07 What I Built / Worked On

  • Product structure
    Project/Search model understanding and project-based lead research flow.
  • Result flow
    Search result saving, project page output display, and simple UI-level Rails views in ERB/HTML.
  • Anthropic flow
    Anthropic integration flow for explainable company candidates.
  • Needs verification
    API key handling improvements, approve/reject actions, and future lead workflow should be confirmed before public detail is added.

08 Key Product Decisions

  • Project-based structure
    A repeatable project/campaign structure instead of one-off search.
  • Explainable output
    Explainable candidate output instead of a raw company list.
  • Saved results
    Saved results instead of temporary AI response.
  • Human review
    Marketing review step before any automated action.

09 Visual Story

  • Rails MVC flow
    Placeholder: Rails MVC flow for Project and Search.

Real screenshots require permission and must avoid exposing internal or customer data.

10 Current Status

OptiGTM is an internal MVP / internal tool and is actively evolving.

Usage data is not yet available. Production adoption is not claimed.

The repo is private and should not be linked. Any real screenshot requires permission and must avoid exposing internal or customer data.

11 What This Proves

  • Workflow to product loop
    I can translate a business workflow problem into an internal product loop.
  • Practical Rails fluency
    I can work with Rails at practical MVP-building level.
  • AI-assisted internal tool build
    I can use AI coding agents to move from product direction to working internal tool.
  • Scoped collaboration
    I can collaborate with marketing and CTO context without overbuilding.