Submission to AgentHack2026 - SupportIQ

Help Agent Builder
, ,
1

Maestro SupportIQ – AI-Powered Customer Support Orchestration with UiPath Maestro

Hello everyone!

I’m excited to share Maestro SupportIQ, my submission for the UiPath AgentHack 2026.

Maestro SupportIQ is an AI-powered customer support orchestration platform built entirely using UiPath Maestro, Agent Builder, and BPMN. The goal was to demonstrate how multiple specialized AI agents can collaborate within a single end-to-end workflow to automate customer support while maintaining quality, transparency, and continuous learning.

The Problem

Customer support teams spend a significant amount of time on repetitive work:

  • Reading and classifying incoming tickets

  • Identifying the correct team

  • Drafting customer responses

  • Escalating complex issues

  • Performing quality checks

  • Documenting resolutions for future reference

While AI can automate parts of this process, real-world support operations require multiple specialized roles working together rather than relying on a single general-purpose model.

Our Solution

Maestro SupportIQ models an enterprise customer support organization using multiple AI agents orchestrated by UiPath Maestro.

Instead of having one AI responsible for everything, each agent has a focused responsibility and collaborates as part of a larger business process.

The workflow includes:

  • Triage Agent – Understands the customer request and classifies intent, category, severity, sentiment, and confidence.

  • Resolution Agent – Determines the likely root cause, generates resolution steps, drafts a customer response, and calculates confidence.

  • Routing Agent – Routes unresolved tickets to the most appropriate specialist AI agent.

  • Support Specialist Agent – Handles authentication, account access, configuration, and general troubleshooting.

  • Engineering Specialist Agent – Handles bugs, API failures, crashes, and infrastructure issues.

  • Billing Specialist Agent – Resolves billing, subscriptions, invoices, payments, and refunds.

  • Product Specialist Agent – Processes feature requests, product feedback, and enhancement suggestions.

  • Review Agent – Performs a final quality assurance review before the response is delivered.

  • Learning Agent – Converts every resolved ticket into a structured knowledge base article to support continuous organizational learning.

Workflow Overview

Customer Ticket
        │
        ▼
Triage Agent
        │
        ▼
Resolution Agent
        │
        ▼
Can AI Resolve?
        │
   ┌────┴────┐
   │         │
  Yes       No
   │         │
   ▼         ▼
Deliver   Routing Agent
Response         │
                 ▼
      Specialist AI Agent
                 │
                 ▼
          Review Agent
                 │
                 ▼
      Deliver Response
                 │
                 ▼
         Learning Agent
                 │
                 ▼
       Knowledge Base

Workflow

Screenshot 2026-06-30 074803
Screenshot 2026-06-30 0748031368×677 101 KB

Challenges

One of the biggest challenges was designing a multi-agent workflow with consistent data flow between every stage of the process.

I also explored using Action Apps for the human review step. Due to limitations in the preview environment, I had to adapt our workflow to use an AI-based review stage while preserving the intended human-in-the-loop architecture.

Designing specialist agents with clearly defined responsibilities and ensuring smooth routing between them was another interesting challenge that pushed me to think about real enterprise support workflows.

What I Learned

Gained valuable hands-on experience with BPMN-based orchestration, prompt engineering, and designing AI workflows that resemble real enterprise operations.

Future Improvements

There are several exciting directions I’d like to explore:

  • Integration with enterprise CRM and ITSM platforms

  • Retrieval from existing knowledge bases to improve resolution quality

  • SLA monitoring and analytics dashboards

  • Multi-language customer support

  • Similar ticket detection using historical knowledge articles

  • Customer satisfaction prediction and reporting

GitHub Repository

The complete project, including the UiPath solution, BPMN workflow, agent prompts, documentation, and setup instructions, is available here:

GitHub:

Demo

A full demonstration video showcasing the workflow and agent orchestration accompanies the project submission: https://www.youtube.com/watch?v=Swva2rLG6iY

I had a great time building this project and experimenting with UiPath Maestro’s orchestration capabilities. I’d love to hear your feedback, suggestions, or ideas for future improvements.

Thank you for reading! :heart:

2

Great work! This is a well-designed use of UiPath Maestro and AI agents for customer support orchestration. Thanks for sharing, and welcome to the UiPath Community! :rocket:

3

Hello, a very nice submission indeed! Did you also submit it on DevPost?

4

Thank you for the welcome :heart:

5

Thank you for your kind words! Yes, I submitted this on devpost first. I saw a field asking for the email i used to sign up to this forum which was needed for the first time builder track. I also noticed many others made similar posts about their submissions, so I made one too. Are you aware of any other requirements that needs to be fulfilled to be eligible for the first time builder track?

6

You just need to provide the link, that is all, no other posts, only if you want to post, you can, but it is not a requirement.