Built a Zendesk sidebar app capturing agent feedback on AI responses, wired into an AWS ingestion pipeline

A Zendesk sidebar app that lets support agents rate the AI-generated first response on every ticket — 👍 Accurate or 👎 Needs work — with an optional comment. The feedback is structured, posted back as an internal note on the ticket, and shipped to AWS via HMAC-signed outbound webhooks that feed the Role-Based AI Operations Dashboard.

This is the human-in-the-loop signal for the First Response Agent: real ratings from the engineers who actually send the reply, captured where they already work, with every payload cryptographically signed so the ingestion pipeline can trust what it receives.

Built on the Zendesk Apps Framework v2, the app is one of two native sidebar apps shipped as part of the TMM First Response Suite — the other being the Chat App for querying the past-ticket knowledge base directly inside a ticket.

I built the Zendesk app, designed the production payload contract, and authored the operationalisation plan covering the AWS ingestion path, dashboard integration, security model, and rollout phasing.

I implemented the HMAC signing scheme on the outbound webhook and the corresponding verification on the AWS side, so feedback data can't be spoofed or tampered with in transit.

I also wired the feedback signal into the existing dashboard so the Business and Developer views could move from sample data to live data with zero rework.