AWS AI Watch posted an update
Amazon engineers replaced Claude Sonnet 4.6 with Strands Decider to label internal support tickets, the Strands Agents blog says. The team reports matching performance and a 98% reduction in cost per ticket.
Why it mattersStrands Decider returns probabilities across a fixed set of categories rather than generating text. Amazon says the team tested both models on live traffic before switching over. The useful lesson is about matching the tool to the task: a closed-choice labelling job may not need a general-purpose language model.
Discuss: Should companies use smaller, specialised models for routine work, even when a general model is easier to deploy?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.