@thsottiaux@argofowl Right after shipping Astra. At least that is what we tell ourselves all the time. But "Astra" gets replaced with the next thing right after it ships. And we love it.
There is a popular twitterpinion going around that companies should stop building agents and only expose MCPs/CLIs. I agree the old way of clicking through a UI is going away but this take is short sighted and misses nuance.
It is almost impossible to expose the full strength of a product through a flat list of tool calls. The API calls behind a real workflow need careful orchestration. And authz is incredibly hard in a headless interface.
Companies that want to take advantage of this shift need all three:
1. An MCP their users can plug into any agent
2. An in-product agent: a carefully curated experience the company controls, owns, and learns from. It becomes a new way of using the product.
3. The product agent exposed through the MCP. Users get the company's agent through their agent of choice.
Simple requests like "pull up my subscriptions on stripe" go through the MCP tools. Complex, curated experiences go through the company's agent (accessed via MCP as well).
Everything should be reachable through MCP, so your agent can talk to my agent. But expecting a general-purpose agent to orchestrate a complex chain of tool calls in the right order every time, is laughable.
The orchestration is the product. The problem is that most companies haven't cracked that yet... so its easy to say: give me all the tools. That's like saying let me use your REST API directly. We just haven't seen the good agentic experiences yet and making this take the standard means we never will.
@ImSh4yy@ikuznetsov_com Broadly agree. Have you considered a corner case where compute gets incredibly costly because it is used for some purpose that someone else is willing to pay a lot for and you simply can't afford it anymore. Science fiction, sure. But don't know how to rule it out.
If your job involves interacting with people more than interacting with machine, you don't have the best intuition for what is possible with today's models.
The reason you want to read code is you are afraid your model is not doing what you want it to do. That is an alignment problem. Solution is to align; not keep checking.
It is tempting when a direction looks promising to just sprint in that direction. It is easy to forget all the other invariants that still needs to hold. Then the laws of nature remind you you still live in their world.
I'm curious when is the earliest anybody in github realized they are going to be hammered by agents. After or before claude code? There were signs even before.