OpenAI is making progress with AI agents that can use specialized software for complex business tasks, training models to understand business rules, execute multi-step workflows, and verify outputs against requirements.

Looking ahead, the effort will assess the reliability of learning company-specific rules, enable multi-step process execution, and allow AI agents to route requests and take actions across contracts and workflows.

Ironclad focuses on complex contracting workflows, ensuring AI agents grasp company rules, perform multi-step tasks, and verify results against requirements.

GenAI is increasingly embedded in professional life, including legal work, but it cannot replace attorneys and should augment rather than supplant legal judgment.

Key partners for law-related plugins include Thomson Reuters, Harvey, Legora, and iManage, highlighting integration of specialist tools and knowledge.

Bounded authority provides a reliable framework: agents operate within defined limits set by a known principal, ensuring accountability and avoiding irreversible commitments.

In programmatic leasing, AI can organize prior negotiations, forms, playbooks, and client guidance to build a data repository that informs future lease negotiations and speeds revisions.

Over time, AI tools learn client strategies and common legal issues, reducing time to produce revised leases.

A repository of prior negotiations, forms, playbooks, and client guidance within AI platforms helps legal teams leverage past strategies for future leases.

Jason Boehmig’s broader view emphasizes ecosystem collaboration and collective progress in AI development.

The partnership traces back to 2023, when Ironclad launched AI Assist on GPT-4 and integrated contract data into ChatGPT.

Overall, the piece points to a shift toward embedding legal domain expertise in AI development, with dedicated legal teams guiding model capabilities and evaluation.