Joerg Hiller
Mar 09, 2026 16:00
Harvey AI unveils Agent Builder, enabling legal teams to create custom AI agents that handle multi-step tasks autonomously. Platform now processes 400K+ daily queries.
Harvey AI has rolled out Agent Builder, a significant upgrade to its workflow automation capabilities that lets legal teams create custom AI agents capable of handling complex, multi-step tasks with minimal supervision. The move comes just weeks after the $11 billion startup announced its Microsoft 365 Copilot integration.
The numbers behind Harvey’s existing agent infrastructure tell the story of why this matters: 400,000+ agentic queries hit the platform daily, users have extracted over 20 million terms through review tables, and teams have generated 445,000+ reports using the company’s Deep Analysis feature. Legal teams have already built more than 25,000 custom workflows on the predecessor system.
What Agent Builder Actually Does
The core pitch is autonomy without chaos. Agent Builder evolves from Harvey’s existing Workflow Builder, but with a crucial difference—these agents can reason through tasks rather than just execute predetermined steps.
A due diligence team, for instance, can deploy an agent that searches documents, builds review tables flagging material risks, and automatically generates a post-closing checklist. Litigation teams can run document reviews that flow directly into first-draft memos. The agent handles the sequencing; lawyers handle the judgment calls.
Harvey has built in human-in-the-loop checkpoints as a core feature. Agents surface decisions and flag moments where user input would improve results before proceeding. An in-house team reviewing regulatory updates might have an agent search policies, flag non-compliant language, then pause to confirm the approach before editing documents directly.
Background Scheduling and Autonomous Execution
Perhaps the most operationally significant feature: scheduled background execution. Teams can set agents to monitor contract expiration dates or run compliance checks on a recurring basis without manual triggering. The agent summarizes findings independently, freeing lawyers from the tedious identification work.
GSK Stockmann has already used Harvey’s agent capabilities to accelerate due diligence reviews. Ashurst reports saving hours on lease summaries through practice-group-specific workflows.
Strategic Context
Harvey’s February 2026 funding round valued the company at $11 billion—up from $8 billion in December 2025 and $5 billion just eight months before that. The trajectory reflects enterprise appetite for AI that can actually execute legal work, not just assist with it.
The company indicated it’s developing “long-horizon agents” for complex tasks like fund formation, plus tailored agents within its Shared Spaces collaboration feature. Existing Workflow Builder users retain full functionality with no migration required.
For legal tech watchers, Agent Builder represents Harvey’s bet that the next phase isn’t better chatbots—it’s AI that can run entire workstreams while lawyers focus on strategy and client relationships.
Image source: Shutterstock