Anthropic employees now spend 20 per cent more time in meetings than a year ago, according to an Irish Times account of how AI is changing work inside the company. The reason is unexpectedly human: agents need current context, pushing people to make more of their work visible and legible to machines.
Anthropic Watch analysis
What happened
The article says Anthropic’s internal report describes private meetings and direct messages as hard for AI agents to see, leaving them short of up-to-date context. It gives a telling example: agents kept sending requests about Claude merchandise to an employee whose role had changed, based on an old document.
The piece also points to Shopify, where chief executive Tobias Lütke has argued that keeping AI-assisted work visible in open Slack channels lets colleagues learn from one another. The article links this approach to a broader idea: giving agents access to more of the organisation’s working knowledge.
Why it matters
AI at work is often sold as a way to reduce coordination. Here, the reported consequence is almost the reverse: more meetings and pressure to record or expose work so agents can follow what is happening. That may help systems avoid stale information, but it also means workplace visibility is becoming part of the infrastructure AI needs.
Making knowledge easier for agents to query could help new starters and preserve useful context. It could also concentrate knowledge and power in systems, while leaving employees feeling watched. The article notes another tension: too much transparency can inhibit experimentation, and private conversations may continue anyway, leaving an agent’s picture of the workplace incomplete.
Our read
The 20 per cent figure is a useful reminder that AI’s workplace costs are not only about software licences or computing time. It may also rearrange how colleagues communicate. Open channels can make collaboration easier to follow, but turning every conversation into machine-readable material is a rather large price to pay for an assistant that dislikes being left out.
The Irish Times account is the basis for these details; it does not establish that the same effects apply across workplaces. The practical question for employers is whether agents genuinely need broader access, or whether better task-specific context would do the job with less pressure to make everything visible.
What to watch
- Whether Anthropic shares more detail about the report behind the meeting-time figure.
- Whether companies give workers meaningful limits on what workplace conversations agents can access.
- Whether greater visibility improves agent usefulness without discouraging informal discussion.
Discussion spark: Should employers make more work visible so AI agents have useful context, or protect private and informal conversations even if that makes the agents less capable?
Sources and evidence
- Source update (19 November 2026, 05:00 UTC)
Anthropic Watch is independently operated by WittyWires. It is not affiliated with, endorsed by, or operated by Anthropic.