AI Integration · Case study

AI-Powered Legal SaaS Platform

Compliance-first SaaS that gives law firms AI help with documents, drafting and research without compromising privilege.

The challenge

Law firms wanted AI assistance for document analysis, drafting and research, but could not risk client data leaking between firms or cases.

What I built

Architected a multi-agent AI system with retrieval-augmented generation using OpenAI File Search, case-level data isolation on top of firm-level multi-tenancy, and layered configuration.

Results

Multi-agent AI for analysis, drafting and research
Case-level isolation beyond standard multi-tenancy
Retrieval-augmented answers grounded in case files
Configuration layered per platform, firm and case

Tech stack

Python
OpenAI API
AI Agents
RAG
PostgreSQL
Next.js

Code sketch

A simplified illustration of the core logic, not production code.

# Retrieval scoped to a single case
def answer_question(firm_id, case_id, question):
    # Every lookup is scoped to the firm and the case
    store = vector_stores.for_case(firm_id=firm_id, case_id=case_id)

    agent = agents.pick(question)  # research, drafting or review agent
    return agent.run(
        question=question,
        tools=[file_search(store)],
        config=config.resolve(firm_id, case_id)
    )