James Ding
Mar 16, 2026 17:51
Harvey CEO announces long-horizon agents, Intapp partnership for ethical walls, and expanded Oxford program as legal AI moves from experimentation to infrastructure.
Harvey AI is positioning itself as foundational legal infrastructure rather than just another productivity tool, CEO Winston Weinberg declared at Harvey FORUM in London this week. The announcement comes as legal AI adoption has surged from 19% in 2023 to 79% of professionals using it in their practice.
The company unveiled three major developments: an expanded partnership with Oxford University’s Blavatnik School of Government, long-horizon agents capable of multi-step legal workflows, and an Intapp integration embedding ethical walls directly into the platform.
From Minutes to Weeks Compressed
The most striking product announcement centers on what Harvey calls long-horizon agents. These systems execute complex legal workflows end-to-end rather than handling isolated tasks. In one demonstrated case, a fund formation reconciliation process that typically takes weeks was compressed to minutes through parallelized agents working across the matter simultaneously.
This capability ties into Harvey’s Shared Spaces feature, where matter-specific environments house all relevant documents and communications. Legal teams can effectively “chat with the matter itself” rather than querying a general-purpose assistant.
Training the Next Generation
The Oxford partnership addresses a growing industry concern. As AI handles more routine work, junior lawyers risk missing foundational training experiences. Harvey’s expanded Law Schools program will focus on AI literacy, output validation, and strategic deployment from day one of legal education.
The timing matters. Industry observers have warned that firms failing to properly train juniors while relying on AI could become “a dying breed.” McCann FitzGerald’s recent enterprise-wide AI rollout explicitly called effective AI use “integral to maintaining position” in the market.
Governance Catches Up
The Intapp partnership tackles the compliance question head-on. As AI becomes embedded in live matters, ethical walls and matter-based permissions must be baked into the system architecture rather than bolted on afterward. For firms handling sensitive client work across competing interests, this integration addresses a critical adoption barrier.
The broader industry shift is clear. GenAI usage among legal professionals nearly doubled from 14% to 26% between 2024 and 2026, with 54% now using AI for drafting correspondence alone. But the focus is moving beyond raw model capability toward what industry analysts call the “architecture of trust” – ensuring AI has complete context while maintaining transparency and auditability.
Harvey’s infrastructure play positions the company for the next phase of legal AI adoption. The question for competing platforms and law firms alike: build similar capabilities internally, or risk watching Harvey define how legal work gets structured for the next decade.
Image source: Shutterstock
