Relevance with Confidence
Relevance with Confidence
Use this template for repeatable, defensible AI review in Claira.
TODO: Screenshot - template picker with Relevance with Confidence selected and a mapped output field.
What this prompt is for
This template helps standardize outputs so reviewers can validate and code results quickly.
Step-by-step in Claira
- Open Prompting > Template picker in your case.
- Select Relevance with Confidence and map the output to one or more review fields.
- Run a 10-25 document sample and inspect edge cases first.
- Adjust instructions outside the prompt body (field mapping, workflow settings), then rerun.
The prompt
Assess whether this document is relevant to the following case context:\n[insert case description]\n\nReturn ONLY one searchable relevance value:\nYESREL or NOTREL.\n\nThen return a confidence score from 1 to 5 (5 = highest) and a concise explanation referencing key language or themes.Recommended customizations
- Add case context before running (date range, custodians, issue framing).
- Keep output constraints explicit (format, allowed values, fallback behavior).
- Version your template configuration so results remain reproducible.
Worked example
Input excerpt
Email dated 2025-02-14 from Dana Lee to Jordan Park confirms the vendor contract amendment was approved and signed the same day.Expected output shape
A concise answer that follows the exact format required by the prompt and is directly mappable to your selected review field(s).Troubleshooting
- If output is too broad, reduce scope and test with a smaller sample.
- If output format drifts, restate strict formatting requirements in workflow instructions.
- If quality varies across models, compare two models on the same fixed sample set.