Identifi Agent

Autonomous Market Intelligence System

2025
R&D / PoCpocNov 1, 2025
LangGraphPythonTavily APIOpenAIVector DB

Identifi started as a Proof of Concept (PoC) to answer a simple question: Can we automate the role of a Market Researcher?

Most "AI Search" tools just summarize the top 3 Google results. Identifi goes deeper. It is designed as an Agentic Workflow using LangGraph.

How It Works

Unlike a linear script, Identifi operates as a graph of stateful nodes:

  1. Planner Node: Analyzes the user's request (e.g., "Find electric motor manufacturers in Southeast Asia") and breaks it down into sub-queries.
  2. Executor Node: Hits the Tavily API to gather raw web data.
  3. Critic Node: Reads the gathered data. If the data is too generic, it loops back to the Executor with refined search terms.
  4. Synthesizer Node: Compiles the validated data into a structured Markdown report.

The Impact

In our pilot with a manufacturing client, Identifi synthesized 50+ market signals per run. It successfully identified 3-5 high-value opportunities that human researchers had missed because the signals were buried in local-language news reports.

This project validated our thesis that the future of search isn't just "finding links"—it's about synthesizing answers.

Gopal Khadka