Route service requests to the right person on a small team
Incoming service requests are read by a decision model that picks which team member should own each one, so requests land with the right person and only unclear ones go to the office manager.
The use case
In a small agency or service business, requests come through one form and someone has to forward each to the right person by skill or account. Jev reads each request along with a short list of team members and what they handle, and picks one owner from that list. Requests with a confidence of 0.85 or higher are assigned directly in the task tool. Lower-confidence requests go to the office manager to assign. Jev does not reply to the client or promise any timeline. The assigned person still reads the request, decides how to handle it, and can hand it back if it was routed wrongly, which is logged to improve the descriptions.
Example workflow
- A request arrives through a website form, email, or Slack Connect channel and is captured by Zapier.
- The request text and the client name are sent to Jev with a list of team members and what each one handles.
- Jev answers a Choice question, who should own this request, with each team member's name plus office manager as the allowed answers.
- Zapier creates a task in Asana or Trello assigned to that person when the confidence clears the threshold.
- Lower-confidence requests are created unassigned and the office manager is pinged in Slack to assign them.
- Any task handed back is logged with the correct owner, and the team member descriptions are updated monthly.
Tools
- Jev (TypeSafe AI)
- Zapier
- Asana
- Slack
- Gmail
- Google Sheets
Risks and guardrails
- A request sent to someone on leave sits idle — guardrail: code removes people marked away in a shared calendar from the option list.
- Vague role descriptions lead to wrong routing — guardrail: write each description in terms of request types, and refine it from handed-back tasks.
- Clients may expect a reply because a task was created — guardrail: no message is sent to the client from this step.
Related recipe
Support Ticket and Review Triage with Policy-Cited Draft Replies: An agent tags every helpdesk ticket and public review by intent, urgency, and sentiment, routes it, and drafts a reply that cites your help center. People approve refunds and all public replies.
Want it set up for you? Request a quote.
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- GitHub repo JevTicketRouter (support tickets: Jev answers scoped questions, .NET rules decide the route): github.com/GhrezaKh74/JevTicktRouter
- Zapier MCP (official page): zapier.com/mcp (checked 2026-11-09)
- Asana MCP server (official page): developers.asana.com/docs/using-asanas-mcp-server (checked 2026-11-09)
- Slack MCP server (official page): docs.slack.dev/ai/slack-mcp-server (checked 2026-11-09)
- Gmail (official page): developers.google.com/workspace/gmail (checked 2026-11-19)