Built with Mobileye’s cloud support team
The Problem
Every support team has the dreaded weekly board-hygiene pass: someone reviews every ticket sitting in review or in progress, reads the latest comment, and decides whether it’s done, waiting on the customer, or abandoned in a shared queue. Then they nudge colleagues about stale items – without flooding them with notification emails.
Here, this was a manual, ticket-by-ticket sweep done by a human every single week.
The Solution
The board agent replaces the manual audit with a scheduled AI-driven review:
- Reads the last comment on every in-scope ticket, using a lightweight LLM to understand the author’s intent in English or Hebrew.
- Transitions ticket status based on meaning: a comment saying something was deployed moves the ticket to verification; a question aimed at the reporter moves it to pending customer; a question for an external party moves it to a third-party-consult state.
- Reassigns orphaned tickets sitting on shared queue accounts to the engineer who was on rotation when the ticket was created.
- Sends one consolidated digest email per assignee listing everything needing their attention – no notification spam.
Small touches like time-of-day greetings and custom branding make the digests feel like a message from a colleague rather than a bot.
How It Works in Practice
- A scheduled job triggers the agent at the end of the work week.
- The agent pulls all recently created tickets in review or in progress.
- It classifies each ticket’s last comment and applies transitions or reassignments.
- Tickets with no comments or unresolved questions are flagged as stale.
- A single summary email goes to each assignee with flagged items.
- Before going live, the team ran the logic in dry-run mode, generating spreadsheet reports and email previews to verify behavior safely.
The Results
A tedious weekly chore became a hands-off routine. Tickets no longer stall silently in the wrong status. Engineers get a structured weekly digest instead of a wall of notifications. Dry-run verification and audit trails meant zero unintended side effects at rollout.
Why It’s Worth Talking About
AI tackling an unglamorous organizational problem with genuine contextual judgment rather than rigid time-based rules – a strong example of responsible automation backed by real verification.
