Claude
AnthropicLong-context writing and code work with a strong editing instinct.
General autonomous agent that runs multi-step tasks in its own environment.
Research and production tasks that can be described and handed over
Work needing judgement mid-task or access to systems with real consequences
An autonomous agent that plans, browses, and produces artefacts in its own workspace. The useful question is not capability but scope: the tasks that work are describable, verifiable, and low-stakes — research memos, comparison sheets, drafts. Anything with an external side effect needs a checkpoint the product does not give you.
Long-context writing and code work with a strong editing instinct.
The assistant inside WhatsApp, Messenger, and Instagram.
One subscription to reach many models at once.
3 alternatives overlap with this one. Compare them →
Pulls LangSmith traces and runs: inputs, outputs, latency, and errors for a project. It turns 'the agent is flaky' into a specific failing step, which is the only wa…
Turn one question into three queries that cover different phrasings.
A curated set of specialist agents and their skills, organised by domain.
Hand a task to another agent CLI, review the diff, land it yourself.