Claude on Amazon Bedrock can now return strict JSON
In brief
Amazon Web Services says Amazon Bedrock now supports structured outputs for all Anthropic Claude models where those models are available. The service can force the model to generate JSON that follows a schema you supply, instead of free text. According to AWS, that removes the need for custom parsing and retries after invalid answers, and can reduce generated text and cost on high-volume jobs.
Source : AWS Machine Learning Blog - « Structured outputs for Anthropic Claude models on Amazon Bedrock »
What it changes for an SME
This change only affects you if you already call Claude through Amazon Bedrock, or plan to. For those flows, answers can come back as valid JSON with the fields and types you asked for, so staff spend less time fixing broken extracts and failed automations. AWS also says limiting the answer to your fields can reduce how much text is generated, which may lower cost at high volume. It does not change your GDPR or EU AI Act duties.
What you can do
- Ask your IT team whether any live AI process already uses Claude on Amazon Bedrock.
- List internal tasks that today copy values from model text into a form, CRM or ticket system.
- If you use Bedrock, check with your provider whether those tasks can send a JSON schema instead of parsing free text.
Let's talk about the time you could save.
In 20 minutes, we find where AI could save you time or money. Free, no commitment.