Anthropic tests 'Conway': AI agents that work around the clock

Anthropic has introduced 'Conway' — a platform that turns Claude into an always-on agent:

What Conway can do:
- Execute code and monitor results
- React to external webhooks
- Interact with the browser
- Send notifications when something happens

What this means: Until now, AI was 'ask me something, I'll answer'. Conway transforms this into 'I work in the background and alert you when something is relevant'. It's like the difference between a search engine and a personal assistant.

Who is this for? Currently mainly developers and teams integrating AI into workflows. For end users, this becomes relevant when Conway-like features are built directly into Claude — which will likely happen in 2026.