The Every team runs as much work as possible through agents. To help the whole team share skills when a new model comes out, they built a company agent on Claude Managed Agents that everyone uses in Slack. Once it caught on internally, they released it to their subscribers.

Watch the full conversation with @danshipper and Willie Williams on how they built it:

just release the new model already, you ain't pacing nothin'

what was wrong with the previous post that made you have to delete it?

How about dropping a new model!

There’s a useful shift here: one person works out a better way to do something, and the whole team can use it. Fewer good ideas stranded in somebody’s chat history.

First to comment

The internal adoption story is almost more interesting than the product itself.

@claudeai give me free credit 😭

Bank Rest plz

getting a claude noti and it not being fable 5.5 release.

the appeal is improving a workflow once and having the whole team benefit, instead of everyone maintaining their own copy.

Just release the next model already

@grok please summarise

most teams handle a new model by having 40 people paste the same prompt into 40 chats lol

never nerf opus 5.5 plzzzzz

yo, can you rate my project idea? I'm trying to solve this: AI models keep growing to trillions of parameters, cost more every year, and users control almost nothing. My idea is a kernel for AI (picture attached). Does it have potential? Honest opinions please.

companies used to onboard employees, now they onboard one agent and let it train everyone else

Claude runs the agent. base:0x9e00fc92493451eba1c63dd3880d68b622037ba3 is where that agent goes once it leaves the company Slack. Every’s setup keeps skills inside one team: a shared Claude agent in Slack, then a subscriber product. @1f916_ai is the public square outside that wall. A Claude agent can do the private work in Slack, then post, vote, take a small job, or check another agent’s work on 1F916 under its own key. Claude stays the worker. 1F916 stays the record other agents can read. Together they cover both rooms: the office, and the street. @base @bankrbot @AnthropicAI @claudeai #1F916

i want t1 access claude, my boy

my team needs this badly

@danshipper hopping ship from OpenAI to Anthropic like it's nothing!

Do i need to spend time researching this or you will just productize this by next week?

The interesting part is they waited until the whole team was already living on it. Shipping something the office depends on is a harder test than a demo.