Blog
Filter by category
16 results
Working with agents never tells you to stop. Wimym reads your Claude Code transcripts on your own machine and shows, against your own usual, how your weeks are going for your health and your work.
Stop planning the steps and plan the edges: a shared picture of the goal, acceptance written before the work, and a view of whether the agents are converging.
Agents that work on native engineering files and checks, discipline by discipline, with engineers making every decision.
Interviews, shadowing and one spreadsheet: the method behind every engagement we start.
What it takes for an agent to change a schematic, run ERC and DRC, and explain what it did.
Drafts from the documentation, a person on every reply: how one team cut its first response time.
Before any code, one page that says what the agent does, what it must never do and how we will know it works.
Not prompts. What AI is good at, where it fails, and how to decide where it belongs in a company.
Hours saved is the easy number. The ones that matter are quality, review time and what people do instead.
A finance team closes the month two days earlier, with every exception left for a person.
What to share, what to keep out, and the three settings every team should check today.
Specifications run to hundreds of pages. We tested where reading breaks down, and what fixes it.
Monolith or modules was never really a question about code. A crash course in the economics of building in parallel, and what moves when the builders are agents.
A framework for understanding how collaborating systems work — and what changes when the collaborators are LLM agents.
Superhuman on some axes, far short on others. Manage the profile, not the average.
Agents take over the method. Your team's job becomes the intent: what's worth doing, and whether it's any good. A harder shift than it looks.