Harvey

@harvey

AI for the world’s most complex legal work.

harvey.ai

Joined March 2023

Following

16.2K

Followers

Affiliates Replies Articles Media

Harvey
@harvey
9h
We post-trained a model for Harvey's Review Tables with @appliedcompute, reducing costs by 50% while improving answer and citation quality. Review Tables allow lawyers to upload up to 10,000 documents and ask up to 500 questions over each document - yielding up to 5 million
Vasudha Rengarajan
@vtrengarajan
10h
Article
Training Frontier Review Table Models with Applied Compute
Authors: @vtrengarajan, Karl de la Roche, @srice120, @nikogrupen, @itsjuliopereyra, @gabepereyra, @caenopy, @jacob_dphillips, @rhythmrg Applied Compute and Harvey partnered to train a model for...
Harvey
@harvey
14h
.@GabrielMacht sat down with law students, lawyers, and leaders at law firms and in-house teams to talk about how they use Harvey. Here’s what they had to say.
00:00
Harvey
@harvey
Aug 12
Harvey for Outlook just got a major upgrade. You can now: - Draft replies grounded in the matter or requests - Search and ask questions over your inbox in plain language, and get answers with citations - Know which emails need attention first and much more.
00:00
Harvey
@harvey
Aug 11
.@gabepereyra gave a talk to @sequoia founders about Harvey’s approach to research. Gabe shared our research playbook, including: 1) Building Legal Agent Bench, our open source benchmark 2) Leveraging the frontier ecosystem to scale post-training and experimentation with a
29:20
Pat Grady
@gradypb
Aug 11
Want world class research capabilities, but don’t have the resources of a big lab? At our recent Sovereign AI event, @gabepereyra shared @harvey ’s “moneyball” approach. Here’s the playbook: 00:00 Introduction 00:37 Building a research lab on a budget 02:28 Legal Agent Bench,
Harvey
@harvey
Aug 11
We post-trained @NVIDIAAI Nemotron 3.5 Lightning on Legal Agent Bench with @trajectorylabs. Here's what we found: 1) Post-training improved agent performance from 0% to 8.3% on held-out LAB tasks, beating both Opus 4.6 and the much larger post-trained Nemotron 3 Ultra. 2)
Trajectory
@trajectorylabs
Aug 11
Continual learning is a bet that the retraining loop will get cheaper over time. With larger models, you can maybe run this loop once every few weeks. But with smaller models, you can run it nightly, per customer. And it keeps recursing: a model per company, then a model per