Archive — US — techcrunch
Anthropic, Google score win by nabbing OpenAI-backed Harvey as a user
Harvey, a legal AI tool initially backed by OpenAI, will now also use models from Anthropic and Google. This marks a significant shift as it responds to competitive benchmarks, indicating a diversification of legal AI resources to enhance performance in specific tasks.
Key points
- Harvey has switched to using foundation models from Anthropic and Google in addition to OpenAI’s, indicating a competitive move.
- The decision is influenced by Harvey's internal benchmark, BigLaw, which highlighted performance differences among various models.
- Harvey, valued at $3 billion, is actively developing AI agents and conducting performance rankings with input from legal experts.
- While extending its model offerings, Harvey emphasizes that it still values its relationship with OpenAI.
Why it matters
- This transition mirrors broader trends in IT, emphasizing multi-cloud strategies and the growing importance of model diversity in AI.
- Increasing competition in the AI space by startups could shift market dynamics by 2025, similar to movements seen during previous tech booms.
- Benchmarking and performance evaluation in AI is becoming crucial, echoing ongoing advancements in machine learning techniques.
Harvey's diversification towards multiple AI models highlights the evolving landscape of legal technology, suggesting that competitive pressures will drive further innovation and excellence in AI applications.