Back to catalog / Content automation
Classify support tickets with an explanation
Classify clear cases using category definitions, then evaluate ambiguous cases with retrieval context and route them for review.
Goal
Generate structured category candidates and reviewable reasoning for a support queue.
Steps
- Define the notebook’s ten ticket categories and boundary rules, then create representative labeled cases. [classify]
- Generate category and explanation with Claude; supply retrieved policy excerpts only for ambiguous cases. [classify]
- Compare prompt versions by per-class precision/recall on a holdout set and route low-confidence/high-impact cases to humans. [classify]
How to verify
Confusion matrix and per-class errors are reviewable; escalation thresholds are explicit application settings.
Troubleshooting
If a class has low recall, inspect overlapping definitions and rare examples; do not report only overall accuracy.
Permissions
Prerequisites
- Python 3.11+, Anthropic API key; a Voyage key only if rebuilding optional embeddings.
Permissions
- Redact production data; send low-confidence and high-impact tickets for human review.
Limitations
The notebook uses synthetic cases; do not automate insurance payments/denials or final customer decisions.
Pinned upstream source
- Repository
- anthropics/anthropic-cookbook
- File path
- capabilities/classification/guide.ipynb
- Pinned commit
9c1e493ddd2b82df399d2a61b9c806c539e18be6- Source checked
- 2026-10-14 · Reviewed against the pinned upstream source
Repository metrics
These values describe the upstream repository, not popularity or likes for an individual skill or playbook.
- Stars
- Not collected
- Forks
- Not collected
- Observed at
- Not collected
No metrics snapshot is available yet.
License: MIT
Main topic
Turn ecommerce campaign materials into a reviewable weekly brief