I got the push in for this right at midnight. Unfortunately, I have had a lot of issues with the class resource group while trying to catch up this week. I was not able to run my usual local quality checks because I ran out of time while trying to get the user stories in. I probably could have merged the user stories and the final project together as one pull request, but I wanted them to feel separate. There may still be bugs, but I did the best with the time I had and I am proud that I was able to put something out there for you to take a look at.

Now on to the technical side of things.

For the final project, I focused on five Azure services that were not part of my user stories version. I tried to make them fit into the existing DeliveryBot system instead of adding services just to check boxes.

Key Vault
Used to store the Azure OpenAI key for the Agent Service, so that secret is not sitting directly in normal app configuration.

Blob Storage
Used to save chat transcripts and support escalation records. This gives the project a simple archive of what the assistant handled.

Azure AI Search
Used as a small knowledge base for the assistant. This helps the chatbot answer from delivery and support information instead of only relying on the model.

Service Bus
Used for support escalations. If a customer has a late delivery, refund issue, cancellation, or similar problem, the assistant can send that as follow-up work.

API Management
Used as one front door for the services the frontend needs. This gives the deployed site a cleaner way to reach the order service, agent service, and bot API.

Overall, the goal was to make the chatbot feel more like a real delivery support assistant. The existing order and robot flow still stays in place, but the final project adds better secret handling, archived conversations, support follow-up, and a cleaner API entry point.