preloader
image

Gemini LangGraph LLM Agent

Project goals were to enhance and extend a document‐review API for a law-firm due diligence workflow, adding dynamic verification checks, external ABN/ACN validation, and configurable recommendations—all powered by a MultiAgent AI pipeline.

Project Details

In this phase, I architected and implemented a LangChain + LangGraph system backed by Google’s Gemini LLM to:

  • Add a Document Verification Section: Introduce AI-driven checks for cross‐references, page numbers, party addresses, signature blocks, and ABN/ACN presence and formatting.
  • Integrate External Validation: Leverage an ABN/ACN verification API to confirm the authenticity and digit‐count correctness of Australian business identifiers.
  • Enable Dynamic Questioning: Refactor the API to accept arbitrary lists of questions and variables from the web app—no more hard-coded prompts—so the AI can answer any set of user-provided checks.
  • Customize Improvement Recommendations: Replace the fixed 10-item recommendation array with a parameterized response, allowing the client to specify how many suggestions they need.
  • Support Long Documents: Use a MultiAgent workflow with Gemini’s extended context window to process lengthy contracts and due-diligence materials end-to-end.
  • Automate Report Generation: Produce ready-to-use Word-formatted AI review reports that seamlessly integrate into the firm’s internal processes.
  • Comprehensive documentation and a replication package ensure the system can be maintained and extended by future developers.
  • Date

    06 Aug, 2024
  • Categories

    Deep Learning, Machine Learning, Generative Ai, Llm, Langchain, Langgraph
  • Client

    Alan Arnott