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