Research
What we are learning about making cloud and local AI work as one.
Start with the manifesto →- Delegation runs on proofA visible call is not a delivery, and a correct answer is not a proven one. For a cloud agent to accept a local agent's work, the result has to carry literal, located and sufficient evidence and a record from the executor. That points to a way to split the work: delegate what is cheaper to check than to do.
- Qwen3.8-27B: new agent features and benchmark gainsQwen3.8-27B adds stronger multi-step execution, controllable reasoning, preserved thinking, native image and video input, a 262K context window, and major reported benchmark gains over Qwen3.6-27B.
- Local AI should not replace your agentThe useful role for a local model is not to take over the workflow. It is to carry bounded work while the agent you already use keeps planning, permissions, and review.