Product
Training a model on your production workflows
Bring executable workflow tasks into Model Factory, improve tool use with environment feedback, and deploy a model your organization owns.
Read the storyResearch, engineering, and ideas from ACTAVA.
Product
Bring executable workflow tasks into Model Factory, improve tool use with environment feedback, and deploy a model your organization owns.
Read the storyEngineering
Production trajectories become simulation environments and a task map of real work, supplying executable tasks for benchmarking and model training.
Read the storyResearch
Evaluation traces guide Cura 1T’s training, from errors in referral tool calls to gaps in clinical responses.
Read the story11 articles
Engineering3 min read
Production trajectories become simulation environments and a task map of real work, supplying executable tasks for benchmarking and model training.
Product3 min read
Bring executable workflow tasks into Model Factory, improve tool use with environment feedback, and deploy a model your organization owns.
Research3 min read
Build tool-use training examples around executable tasks, then verify the tool calls and the resulting application state.
Research3 min read
Train agents to ask for missing information and use the answer to complete a task.
Engineering3 min read
Agents share files and progress in Workspace. People review proposed actions, request changes, and decide when work can continue.
Product3 min read
An imaging referral shows how to build an agent with a clear task, the right tools, and human review before submission.
Company3 min read
Sanmi Koyejo explains why healthcare AI evaluation starts with the work clinicians do and follows an agent through the full workflow.
Company2 min read
A conversation with Bo Li about building security into the tools, permissions, and review decisions that govern an agent’s actions.
Company2 min read
Our OpenAI partnership brings frontier models into workflows with connected tools, task-level evaluation, and human review.
Bring one workflow your team still runs by hand. We'll show you the agent that runs it.
Research, engineering, and ideas from ACTAVA.
Product
Bring executable workflow tasks into Model Factory, improve tool use with environment feedback, and deploy a model your organization owns.
Read the storyEngineering
Production trajectories become simulation environments and a task map of real work, supplying executable tasks for benchmarking and model training.
Read the storyResearch
Evaluation traces guide Cura 1T’s training, from errors in referral tool calls to gaps in clinical responses.
Read the story11 articles
Engineering3 min read
Production trajectories become simulation environments and a task map of real work, supplying executable tasks for benchmarking and model training.
Product3 min read
Bring executable workflow tasks into Model Factory, improve tool use with environment feedback, and deploy a model your organization owns.
Research3 min read
Build tool-use training examples around executable tasks, then verify the tool calls and the resulting application state.
Research3 min read
Train agents to ask for missing information and use the answer to complete a task.
Engineering3 min read
Agents share files and progress in Workspace. People review proposed actions, request changes, and decide when work can continue.
Product3 min read
An imaging referral shows how to build an agent with a clear task, the right tools, and human review before submission.
Company3 min read
Sanmi Koyejo explains why healthcare AI evaluation starts with the work clinicians do and follows an agent through the full workflow.
Company2 min read
A conversation with Bo Li about building security into the tools, permissions, and review decisions that govern an agent’s actions.
Company2 min read
Our OpenAI partnership brings frontier models into workflows with connected tools, task-level evaluation, and human review.