DeepJudge launches open protocol for AI context handoff

DeepJudge on Aug. 13 introduced Agent Handoff Protocol, an open standard meant to let users move context between AI tools without starting over. Harvey is integrating the protocol in beta this month, while Thomson Reuters said it is supporting the effort and will share implementation details later.

Why it matters: - AI users often lose the thread when they switch between products, forcing them to rebuild prompts, upload materials again and recreate prior analysis. - DeepJudge is positioning Agent Handoff Protocol as a way to preserve context across specialized AI systems, which could matter most in high-stakes legal and professional workflows. - The protocol could make it easier for legal teams to use different AI tools for different tasks without sacrificing continuity.

What happened: - DeepJudge introduced Agent Handoff Protocol on Aug. 13 in Zurich. - The protocol is designed to transfer a user and the context needed to continue work from one AI system to another. - Harvey announced an integration of Agent Handoff Protocol that is entering beta this month. - Thomson Reuters said it is supporting Agent Handoff Protocol and will provide more details on its implementation and timing in future announcements.

The details: - DeepJudge built the protocol and is releasing it as an open protocol. - The specification and implementation resources are available on GitHub. - The protocol is meant to carry objectives, supporting materials, conversation history and prior work between systems. - A lawyer researching a precedent question in DeepJudge could move into Harvey with the relevant context already in place. - The workflow would avoid re-uploading materials, recreating prompts or rebuilding prior analysis. - Yannic Kilcher, DeepJudge co-founder and CTO, said no single AI experience will fit every task and users should be able to move their context to the best tool for the job. - Joel Hron, Thomson Reuters chief technology officer, said professionals will increasingly work across specialized AI experiences and those systems need to work together without losing context, trust and expertise. - Siva Gurumurthy, Harvey chief technology officer, said the most capable legal workflows will use more than one specialized system and that Harvey is glad to be among the first to implement the protocol. - DeepJudge says it wants other AI providers, specialist platforms and law firms building in-house AI tools to adopt the protocol and help shape its development.

Between the lines: - DeepJudge is treating interoperability as the product story, not just another feature launch. - Openly releasing the protocol could help build a shared standard before competing vendors lock in separate workflows. - The early support from Harvey and Thomson Reuters gives the protocol immediate credibility in legal AI, where trust and continuity matter.

What's next: - Harvey’s beta integration is set to begin this month. - Thomson Reuters will share more information on its implementation and rollout later. - DeepJudge is inviting broader adoption from AI vendors and law firms, which will determine whether the protocol gains traction as a common standard. - The protocol’s future will likely depend on whether enough platforms adopt it to make cross-tool handoffs routine. - DeepJudge is likely to continue expanding the specification and implementation guidance as more partners join.