Harmony - AI agents that resolve IT/HR tickets inside Slack and Teams
by•
Harmony is an AI-native ITSM platform for IT teams. Instead of bolting a chatbot onto a ticketing tool, we built the helpdesk around AI agents: employees ask in Slack or Teams, and 100+ production-ready agents handle password resets, app access, onboarding, device issues and more end to end, escalating to a human only when needed. Ticketing, asset inventory, SaaS management and workflows live in one workspace, so agents have context to act. Teams hit 60%+ auto-resolution in days.
Replies
@ranrib Congrats Ran and team! Love that you’ve taken this all the way from managing tickets to agents actually resolving requests and closing the loop. There’s so much further this model can go as more of IT operations becomes autonomous.
Curious, where do you draw the line today between what an agent can do autonomously and what still needs human approval?
I’m obviously biased because I work at Harmony 😊 but as the person leading Operations here, I also get to experience the product as a user every single day.
Ops is made up of small, repetitive requests: onboarding, offboarding, access, permissions, apps, and a million other things that come up during the day. Having Harmony take care of so many of these directly in Slack saves me a lot of time (and a lot of back and forth 😅).
Very proud to be part of this team and to actually use what we’re building every day. Congrats team! 🚀
Congrats!
Lfg! What a great idea team love it
Congrats on the launch keep shipping have some ideas in mind would love to connect
Love it! looks super interesting, and a true problem space!
IT teams have always struggled, from experience back in the old days :)
Best of luck!
Harmony dev here, so pinch of salt. Everyone asks about the agents, but most of the hard part was identity. The person asking in Slack has to be provably the same human as the Okta user, the Jamf device owner and the HiBob record, and their name is spelled three different ways across those. Get that right and the agent part almost falls out. Get it wrong and you've built something that confidently resets the wrong person's password.
As VP of R&D at Harmony, what stands out to me isn't only the 60%+ auto-resolution number, it's the architecture underneath it: the model never executes actions directly, it only matches requests to pre-approved workflows, and every object it references has to resolve against your real IdP/MDM data or the request just fails. That's the right answer to "what happens when the agent is wrong or fed a malicious prompt," which is the question every security team actually asks and most AI tools dodge. Scoped credentials, eligibility rules, human approval on sensitive actions, and a full audit trail mean prompt injection can steer the model's interpretation but has nowhere to steer it to. That's what building agentic AI the way a security team would've designed it themselves looks like! 🚀
The hardest part of building internal tools isn't the model, it’s getting reliable context out of your corporate tools. If your HR suite, MDM, and directory services don't talk to each other seamlessly, your AI is operating blind. Once your data layer is clean and accessible with Harmony, adding automation takes minimal effort.
the "60%+ auto-resolution within the first couple of weeks" number is the claim I'd want to poke at, that's a fast ramp for a helpdesk tool since most of the friction in IT support isn't the request itself, it's edge cases where the AI has to know when to escalate instead of confidently doing the wrong thing. what does the failure mode look like when an agent gets a request wrong, does it just get flagged by the employee or is there a way it catches its own uncertainty before acting on something like access changes?
agents that actually close IT tickets not just chat about them 👏 whats the false-positive rate looking like?