Stop counting tokens, start counting what ships
When a developer gives an AI agent a coding task, the agent does more than answer once.
When a developer gives an AI agent a coding task, the agent does more than answer once.
The short version
- It plans, generates code, runs tests, and repeats the process to correct errors.
- A single task can use 10 to 50 times more tokens than a basic chat prompt.
- The more useful the agent becomes, the more calls…
What happened
Why rationing agentic coding backfires, and what to measure instead. The more useful the agent becomes, the more calls it makes.
Why it matters
With per-token pricing, the workflows creating the most value can also become the most expensive.
Summary by Nerd News Network. Read the full article at MIT Technology Review — AI via the links above and below.