Automating 90% of Financial and Legal Work with Agents

Financial and legal professionals spend enormous amounts of time reviewing vast market research, contracts, and regulatory documents. To address this, Hebbia has introduced Matrix, a multi-agent AI platform that handles complex workflows end to end.

Rather than relying on a single model, Matrix orchestrates multiple AI agents in parallel by simultaneously leveraging OpenAI's o3-mini, o1, and GPT-4o. This overcomes the limitations of conventional RAG (Retrieval-Augmented Generation) approaches and provides an effectively 'unlimited' context window for offline data.

Matrix's key features include:

• Breaking down complex queries into structured analytical steps
• Intelligent routing to the AI model best suited for each task
• Processing full documents rather than excerpts, with transparent citations
• Proactive updates through self-improving index construction

In terms of performance, Hebbia combined Matrix with the o1 model to significantly boost accuracy on complex legal and financial document benchmarks, improving from 68% with conventional RAG to 92%.

Real-world adoption results are notable as well. Investment bankers are saving 30 to 40 hours per deal, and law firms have cut credit agreement review time by 75%, reducing legal costs by $2,000 per hour.