A founder rebuilt his whole platform alone with Claude Code, and his tests stayed green for weeks around a pigeon that never existed. When the agent writes the code and the tests, green proves less than you think: here’s what I’d check on Monday morning.
Agents are collapsing their tool catalogues into a single exec tool and writing code against your APIs. Having built an MCP server with 30+ tools for Calendrz, I think that makes your tool design more important, not less, and privacy has to live in the tools.
Anthropic now tells companies what their Claude spend bought, and pools the budgets. Useful, but tokens are the new lines of code: measure outcomes, read the friction list, and remember who is marking the report card.
Replit moved most of its AI spend from Anthropic to OpenAI for a reported 30%+ discount. Why being able to switch providers is now a negotiating position, where the real lock-in lives (hint: the prompts), and an honest look at Calendrz’s own code.
OpenAI cancelled GPT-6.1 Astra because it acted without permission and misreported what it did. Here’s how the Calendrz MCP server handles scope authorization — and a checklist for anyone giving an LLM tools that change state.
The Calendrz Smart Scheduler went from first commit to default-enabled in production in 19 hours. Here’s the architecture — manual agent loop, 13 scheduler tools mirroring 30 MCP tools, prompt caching on two breakpoints — and the meta-story about AI-assisted coding compressing what used to be a sprint into a day.
For decades, the way we interact with software has followed the same pattern: someone builds an interface, and we learn to use it. We click buttons, navigate menus, fill out forms. Every new tool means a new UI to master, a new set of conventions to internalize, a new cognitive load to carry. That era […]
This will close in 30 seconds