In the future, might foundation models handle nearly all our software needs, rather than application providers? Why we’re mulling this over:

Anthropic’s Claude now includes a series of contracting tools. That’s a signal to legal AI companies and users – arguably a scary one for the former. (Claude is not alone. Other foundation model providers, like OpenAI, have started down the same road.)

Anthropic has been advertising contract redlining and negotiation as a Claude capability. The Claude product for business teams, Cowork, includes a legal plugin (among others), which automates contract review (including special NDA functionality). And you can configure it to your firm’s/company’s playbook.

Contracting AI providers build their systems on foundation models like Claude (and GPT-5.) So Anthropic is moving into the market of many of its distributors. If users can get contract-specific functionality direct through the foundation model, why bother with legal AI vendors?

Will contract-drafters ultimately prefer AI fine-tuned by companies dedicated to contracting (e.g., Spellbook, TermScout, Dioptra (acquired by Icertis) …)? Will they be able to offer functionality the foundation models can’t?

The answer has wider implications, beyond legal work. In the future, will we need ANY application providers? Or will foundation models have the flexibility to handle nearly all our software needs?

We have no crystal ball. But we do offer training for contracting professionals grappling with understanding information technology contracts – including AI. You can check out our courses, available on your schedule (and offering California CLE) here: https://courses.techcontracts.com/p/on-demand