Project Story

Every technical support engineer knows the pattern: a user reports a problem, the support team investigates multiple systems, searches documentation, tries different fixes, waits for results, and finally verifies whether the issue is actually resolved.

The challenge is not always finding the answer. The real challenge is turning that answer into action.

We built ResolveAI to explore a simple question:

What if an AI agent could work alongside a support engineer and safely handle the entire incident-resolution process?

Instead of creating another chatbot that only recommends troubleshooting steps, ResolveAI acts as an autonomous IT support engineer.

When an incident arrives, ResolveAI first understands the problem and creates an investigation plan. It then delegates tasks to specialized agents that examine system health, network connectivity, authentication, application status, historical incidents, and technical documentation.

Once it identifies a likely root cause, ResolveAI can select an approved remediation and execute it inside a controlled environment.

But we wanted to solve one more important problem: How does the agent know that it actually fixed the issue?

ResolveAI therefore includes a dedicated Verification Agent. After every remediation, the system performs independent checks to confirm that the original problem has been resolved. If the remediation fails or the agent does not have enough confidence to proceed, it stops and prepares a detailed escalation for a human engineer.

This creates a complete agentic workflow:

Understand → Investigate → Decide → Act → Verify → Resolve or Escalate

The project was inspired by the repetitive, multi-step nature of technical support and the growing need for systems that can do more than generate text.

Our goal was not to replace human support engineers.

Our goal was to give them an intelligent teammate that can handle repetitive investigation and remediation work, provide evidence for its decisions, and know when human judgment is required.

Through ResolveAI, we learned that building a useful autonomous agent is not just about giving an LLM access to tools. It requires thoughtful orchestration, persistent state, controlled actions, verification, observability, and clear boundaries around autonomy.

ResolveAI represents our vision of the next generation of technical support: AI that doesn't simply tell you what to do, but safely does the work—and knows when to ask for help.

What it does

How we built it

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for ResolveAI — Autonomous IT Support Agent

Built With

Share this project:

Updates

posted an update —

Building ResolveAI — Autonomous IT Support Agent

Excited to share our progress on ResolveAI, an agentic AI solution designed to transform the way technical support incidents are investigated and resolved.

Instead of creating another chatbot that simply provides troubleshooting instructions, our goal is to build an AI support teammate that can:

Understand incoming incidents Investigate technical signals Search relevant troubleshooting knowledge Identify probable root causes Execute approved remediation actions Independently verify the result Resolve the incident or escalate it with evidence

Our core workflow is:

Understand → Investigate → Decide → Act → Verify → Resolve or Escalate

The project is being designed around Gemini, Google Agent Development Kit (ADK), and Google Cloud, with a focus on reliable agent orchestration, controlled autonomy, persistent state, and verification.

One of my key learnings so far is that an autonomous agent should not simply say “the problem is fixed.”

It needs to prove it.

That principle is at the heart of ResolveAI.

I am excited to continue building, testing, and refining the prototype for the All Things Agentic Hackathon.

Gemini #GoogleCloud #GoogleADK #AgenticAI #AI #TechnicalSupport #Cloud #Automation #Hackathon #ResolveAI

Log in or sign up for Devpost to join the conversation.

Submission history