The Anti-ChatGPT: Thomson Reuters’ Multi-Agent System Drops 20-Hour Tasks Down to Just 10 Minutes

Catalyst

As part of the latest era of enterprise AI advancement, Thomson Reuters has announced a novel multi-agent system that significantly accelerates complex legal research. Called “Deep Research,” this revolutionary platform is powered by agentic AI to change how work is done in the legal space. Professionals can complete what used to take 20 hours in about 10 minutes.

The Hitchhiker’s Guide to Legal Research

Lawyers struggle with the time-consuming process of research. Reviewing case law, statutes, and administrative rules can take several days, consuming resources that could be used elsewhere.

Thomson Reuters’ Deep Research addresses this challenge with a multi-agent system designed to perform highly complex legal research efficiently.

• Unlike traditional AI systems, Deep Research does not focus on quick answers.
• The platform plans, executes, and analyzes tasks using a comprehensive dataset of over 20 billion documents, including:
  • Fresh case law
  • Statutes
  • Administrative rulings
  • Secondary materials
  • Structured legal content

This extensive dataset ensures that AI agents have access to a broad, accurate knowledge base, resulting in more reliable outputs.

The Role of Multi-Agent Collaboration

Deep Research is built on a multi-agent architecture. Instead of relying on a single AI model, the system uses a constellation of specialized agents that collaborate to solve complex problems.

• Each agent is responsible for a specific task, such as:
  1. Data gathering
  2. Analysis
  3. Synthesis
• This task separation allows parallel processing, accelerating the overall research process.
• Agents interact and coordinate to produce coherent and comprehensive results.
• Benefits include faster research and improved quality due to diverse perspectives and expertise.

Improvement of Legal Accuracy and Error Correction

A standout feature of Deep Research is its ability to reduce errors and “hallucinations”—situations where AI generates plausible but incorrect information.

• Standard AI models, such as retrieval-augmented generation (RAG), are prone to inaccuracies, which is risky in high-stakes legal areas.
• Thomson Reuters addresses this with direct citations from its database, providing official and accurate information.
• This approach mirrors human legal research, offering nuanced insights while reducing discovery time.

“When we hear how law firms are using Westlaw for fast answers, we know the motivation is not just speed but broad perspective to help with understanding difficult, complex legal issues,” explains Mike Dahn, Head of Westlaw Product.

Integration with Westlaw and CoCounsel

Deep Research is part of Westlaw, Thomson Reuters’ premier legal research platform:

• Utilized by over 12,000 law firms, more than 4,000 corporate legal departments, and most top U.S. courts and law schools.
• Integration allows legal professionals to leverage advanced AI within existing workflows, improving productivity without disruption.

Deep Research is also integrated with CoCounsel, an agentic AI system for tax, audit, and accounting professionals:

• Brings agentic AI beyond law
• Automates sophisticated workflows
• Enhances efficiency across industries

This reflects Thomson Reuters’ commitment to advancing AI-powered solutions across multiple professional domains.

A Blueprint for Enterprise AI

The success of Deep Research provides a model for enterprise AI:

• Demonstrates that slowing down AI for deeper analysis can create significant business value
• Highlights the importance of depth and precision alongside speed in AI development
• Shows that AI systems can plan, reason, and act effectively within professional workflows
• Emphasizes trustworthiness and accountability, ensuring AI can reliably perform critical tasks

Conclusion

As AI systems continue to evolve, multi-agent systems like Deep Research offer a meaningful path forward for complex professional tasks.

• Delegating work to specialized agent systems improves efficiency while maintaining high standards of precision and dependability.
• Benefits extend beyond legal and financial sectors to industries such as:
  • Healthcare
  • Education
  • Scientific research

Organizations increasingly see AI as more than automation, but as a solution-oriented collaborator. Multi-agent systems exemplify how AI can work collaboratively, reason systematically, and integrate deeply to solve complex problems efficiently.

Thomson Reuters’ Deep Research sets a new benchmark for enterprise AI, demonstrating the transformative potential of agentic, collaborative AI systems.