Harvey and Thomson Reuters have developed their own AI models, Tenet and Thomson respectively, moving away from reliance on OpenAI and Anthropic. This shift aims to reduce costs and increase control over their AI services.
TL;DR
- Harvey and Thomson Reuters are launching their own AI models, Tenet and Thomson, built on open-source technology.
- This move aims to reduce dependency on OpenAI and Anthropic, potentially lowering costs and increasing profitability.
- The shift comes with new responsibilities, including model maintenance and security, but could improve AI system functionality.
What happened
Harvey announced the development of Tenet, a custom AI model built on the open-source Kimi K3 model. Thomson Reuters revealed plans to roll out its own model, Thomson, based on an unnamed open-source foundation.
Both companies have previously relied on frontier model labs like OpenAI and Anthropic for their AI computing needs. This shift marks a significant change in their service provision strategy.
The move is driven by the potential to reduce costs and increase control over their AI services. However, it also means taking on new responsibilities, such as model maintenance and security.
Why it matters
For developers and startups, this trend highlights the growing importance of open-source AI technology. It also underscores the potential benefits of custom AI models, including cost savings and improved functionality.
For investors, this move could signal a shift in the legal tech industry towards more independent, self-sufficient AI solutions. It also highlights the potential risks and responsibilities associated with developing and maintaining custom AI models.
The impact on end-users, such as attorneys, remains to be seen. While there may not be a noticeable change in the average user experience, the new models could potentially improve the overall functionality of AI systems in the legal field.
Key facts
- Harvey's Tenet model is built on the open-source Kimi K3 model.
- Thomson Reuters' Thomson model is based on an unnamed open-source foundation.
- Both companies have previously relied on OpenAI and Anthropic for their AI computing needs.
- The shift to custom models aims to reduce costs and increase control over AI services.
- New responsibilities include model maintenance and security, according to industry experts.
- The move could potentially improve the functionality of AI systems in the legal field, according to Daniel Linna, a professor at Northwestern University's Pritzker School of Law.
- The average attorney using tools built on the new models might not see much of a change, according to Patrick DiDomenico, a legal technology adviser.
- This trend is part of a broader experimentation wave in the AI industry, including among frontier model labs like OpenAI and Anthropic.
Context
This shift comes amid a broader trend in the AI industry towards custom, independent solutions. As more companies develop their own AI models, the landscape is becoming increasingly diverse and competitive.
The move also reflects a growing interest in open-source AI technology. As more companies leverage open-source models, the potential for innovation and customization is expanding.
However, this trend also highlights the challenges and responsibilities associated with developing and maintaining custom AI models. As companies take on more control over their AI services, they also take on more risk and responsibility.
