Advancing computer use with Ironclad
What happened
OpenAI partnered with Ironclad, a contracting software company, to train AI agents on complex legal and business workflows. They created 11 tasks—like setting up nondisclosure agreements and procurement approval processes—using Ironclad’s platform and evaluated them with detailed criteria. The new GPT‑6 Astra model scored 32% higher and took 48% less time than the previous GPT‑5.6 Sol on these tasks, achieving an average score of 55.0% versus 41.6% for Sol. Researchers used synthetic training tasks and reinforcement learning within Ironclad’s hosted environment to improve the model’s performance.
Why it matters
For ordinary readers, this means AI can handle more intricate business paperwork faster, potentially reducing the time professionals spend on routine contract setup. It also shows AI is moving closer to reliably assisting with real‑world legal and procurement work.
Reality check
An automated, sceptical read of the sourceIncremental. The improvements are notable but build on existing models and focus on a specific partnership rather than a wholly new capability.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from OpenAI
Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026OpenAI launches ChatGPT Wallet, AI agents pay for purchaseOpenAI · 13 Nov 2026OpenAI disbands its robotics team as its general models start running robotsOpenAI · 13 Nov 2026Yang-Mills, OpenAI proof, and the mass gap problemOpenAI · 13 Nov 2026This page was collected from openai.com and summarised automatically by AIPulsee's pipeline. See how it runs.