• One agent beat four not because names confused people, but because the valuable work crosses departments. Departments exist since no one person can do a whole job; an agent that completes work end-to-end doesn't need that split.
• The trigger was engineers running agents in parallel via git worktrees with Claude Code and Codex and getting roughly 5x more done on some tasks, which is what justified a six-person team to spread it company-wide.
• The bottleneck moved from intelligence to context. Frontier models are already smart enough for most business tasks; what they lack is your workflows, your history, and judgment calls absent from any training set.
• Two people wired a data-analyst agent to Slack, GitHub, ClickHouse, Salesforce and PagerDuty over MCP (Model Context Protocol). It investigated customer issues across all five in minutes, work that used to take an afternoon.

• AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
• context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
• MCP — The Model Context Protocol — an open standard that lets any AI assistant plug into any tool or data source without custom integration code.

If you're trying to get your whole company using AI — not just engineers — the biggest trap is building one agent per department. This post explains why Sierra collapsed theirs into a single cross-functional agent and what they learned doing it.

• postAI Makes It Easier to Build the Wrong ProductGrey
• clipComputers-in-the-90s analogy for AI adoptionClaude
• articleBeyond code generation: rethinking engineering productivity in the age of AI agentsIlya Yakovlev,Andrew Cheung,Binoy Dash,Simran Jumani,Dmitriy Meyerzon,Mark Breitenbach,Ishan Mishra,Kazuaki Okumura

Checking sign-in…

Loading comments…