Ravi's Wednesday, two ways

Ravi, a backend developer at a mid-size logistics company.

Ravi starts Wednesday with a ticket for a new returns endpoint. Instead of opening the editor and staring at the route file, he pastes the requirement into Claude and asks for a route structure, request body, and validation notes. The draft is close enough to shape, and he spends the next twenty minutes on the parts the tool could not know: how the warehouse system names a return, and which status codes the mobile app expects.

After lunch, a slow orders query lands on his desk. He pastes the query plan into ChatGPT and asks where the time goes. It points at a missing index on a date column and suggests a rewrite for the join. Ravi checks both against the staging database, keeps the index, and leaves the rewrite for a day when he has more room to test.

At four, a payment webhook starts failing. He drops the last hour of logs into Gemini and asks for a timeline. The summary shows the failures began right after a deploy, and the error traces back to a changed header. He writes the incident note from that summary, adds the fix, and closes the ticket before six.

The time he wins back goes to a walk before dinner and a quiet hour with his daughter. On Thursday he will use the same habit on the old pricing module, asking the tool to explain it first and write tests before he touches a line.