How to classify incoming requests automatically
To classify incoming requests automatically, the system has to read the request and set the fields that determine where it goes. In Jestor, an AI agent can assign area, priority and even sentiment on the fields you authorize, so triage stops being a person's morning.
Why manual triage is the most wasteful step
Someone opens each request, reads enough to categorize it, sets a field and moves on. It requires no expertise and consumes a predictable share of every day.
It is also where the backlog starts, because triage happens once a day while requests arrive continuously.
What automatic classification changes:
- Requests routed within seconds of arriving
- Classification applied consistently rather than by whoever is on duty
- Urgent items surfaced without waiting for a triage round
- The person doing triage freed for actual resolution
- Categories applied uniformly, which makes reporting meaningful
What can be classified
| Attribute | What it determines | Verification |
|---|---|---|
| Area | Which team receives it | Misrouting is visible immediately |
| Priority | Position in the queue | Spot check against outcomes |
| Category | Reporting and patterns | Review the distribution monthly |
| Sentiment | Whether escalation is warranted | Compare against complaints |
| Attachment content | Context from documents | OCR extraction confirmed |
Step by step to set it up
- Define your categories first, keeping the list short enough to be applied consistently.
- Create the agent with instructions describing each category precisely.
- Authorize only the classification fields, not the rest of the record.
- Enable attachment reading, so documents inform the classification.
- Run it alongside manual triage for a week and compare.
- Measure the misrouting rate, which is your real accuracy number.
- Refine the category definitions, which is usually the fix when accuracy is low.
Why the category definitions matter more than the model
When classification is inaccurate, the instinct is to blame the model. More often the categories overlap, so a human triaging the same request would also hesitate. Writing definitions precise enough that two people would agree is what makes automatic classification reliable.
Why choose Jestor
In Jestor, requests arrive natively by email and WhatsApp, becoming records automatically. The agent classifies within authorized fields, reads attachments with OCR, and can move the card to the right queue. Everything it does is recorded, and misrouting is visible on the same board the team already uses.
Frequently Asked Questions
Can the agent read an attached document to classify?
Yes. OCR reading lets the attachment inform the classification. See jestor.com.
What if classification is inaccurate?
Usually the categories overlap. Tighten the definitions before changing anything else.
Can it set priority as well as area?
Yes, along with category and sentiment, on the fields you authorize.
Video Tutorial: Step by Step
With Jestor, you can automate workflows, connect teams, and build internal systems your way, all without code and powered by AI. Discover Jestor at jestor.com and see how to take your company's operations to a new level of efficiency and control.