Skip to main content

Hackathon Spotlight: “Ambient Agents represent a shift from reactive tools to proactive intelligence”

  • October 24, 2025
  • 1 reply
  • 93 views

JenHanley
Forum|alt.badge.img+1

Device issues are one of IT’s biggest headaches, but many can be proactively prevented. 

@mirel.ivan from Keysight built an Ambient AI agent that is able to listen to Nexthink events and alerts device owners the moment a problem arises. It then evaluates the event details, compares them against available Remote Actions, and automatically runs the right one to fix the issue. This agent can powerfully reduce manual triage and helps keep employees productive, rather than troubleshooting.
 

Check out his video here included in this post.

We were able to catch up with Meril after the event and here’s what he had to say:

​​What inspired you to build your AI agent?
We are very excited at Keysight about the new features offered by Moveworks so I was inspired by the potential to simplify complex workflows and make information instantly accessible. Even though we are quite young with our Moveworks AI Assistant deployment, the idea of creating an agent that could proactively assist users, rather than just react, felt like a natural evolution. 

 

What was your biggest “aha!” moment during the hackathon?
My biggest “aha!” moment was realizing how quickly Ambient Agents could integrate with existing systems and deliver real value without heavy customization. Seeing the first end-to-end workflow automated in minutes was a game-changer and it showed me how powerful and accessible this technology really is.

 

What excites you about Ambient Agents?
I think that one of the most powerful aspects is their ability to work quietly in the background, anticipating needs and acting without constant human intervention. They represent a shift from reactive tools to proactive intelligence, which can transform productivity and user experience across organizations.

 

What are you most excited to build next with Ambient Agents?
I’m excited to build agents that handle multi-step processes across multiple platforms like IT troubleshooting or Onboarding workflows, without requiring manual coordination. Adding contextual awareness and personalization will make these agents even more impactful.

 

Can you share how you used Agent Architect, Moveworks Docs, or QuickStart while building your agent?
During the London Hackathon I was introduced to Agent Architect that was invaluable for structuring the agent’s logic and defining triggers. Moveworks Docs provided clear examples and best practices, which helped me avoid common pitfalls. QuickStart were perfect for accelerating development, so I used them as templates to get core functionality running fast.

 

What advice would you give to other teams getting started with Agent Studio?
Start small and iterate quickly. Focus on one high-impact use case first, then expand. Leverage QuickStart as they save time and help you learn by example. And don’t hesitate to experiment; the platform is flexible enough to support creative solutions.

 

Is there anything else you'd like to share with the builder community?
Building with Ambient Agents is not just about automation ... it’s about creating experiences that feel intuitive and human centric. Share your learnings, collaborate, and keep pushing boundaries. The more we innovate together, the more value we deliver to end users.

 

Inspired by Mirel’s agent? Tell us about the agents you’re building in the comments!

 


 

 

1 reply

  • Known Participant
  • January 22, 2026

This is really cool. The proactive piece is what makes it click for me. Instead of waiting for someone to submit a ticket, the agent just fixes the problem before it becomes a problem.

I love that he built this at the hackathon and had that moment of realizing how fast you can wire things up. The Nexthink integration is smart too, especially for device management where you're already sitting on all that telemetry data.

The advice about starting small and iterating quickly is spot on. Way easier to prove value with one solid use case than trying to boil the ocean.

Really cool to see what people are building with this.