Launched this week

OpenFrame
AI agents that run your IT. You approve every fix.
500 followers
AI agents that run your IT. You approve every fix.
500 followers
OpenFrame gives small IT teams two AI agents to handle the work that never ends. Mingo patches devices, fixes issues, and handles routine technical tasks, with every change approved by your admin. Fae answers employee IT requests so your team can focus without constant interruptions. A weekly report shows leadership what got done. Endpoint management, AI agents, and IT support in one platform for $1 per device.










Free Options
Launch Team / Built With

CotypistLocal AI Autocomplete in your voice, anywhere on your Mac
Promoted



Hey Product Hunt, I’m Vlad, Head of Marketing at Flamingo, the team behind OpenFrame 👋
The problem
At most growing companies, IT is one or two people. They patch laptops, onboard new hires, chase security alerts and answer “quick questions” all day. When everything works, nobody notices. When something breaks, it’s on them.
Most tools for this job fall into two camps:
❌ Big IT suites built for large IT departments, and priced and configured like it.
❌ AI chatbots that tell you what to do, then leave you to go do it in five other consoles.
We wanted an AI that does the work, with a human who decides what goes ahead.
What OpenFrame does
OpenFrame is an IT platform with two AI agents built in.
🔷 Mingo does the technical work. Ask in plain language: “patch every laptop that’s behind,” “why does this machine keep crashing.” Mingo investigates, puts together a plan and runs it. Anything risky waits for your approval.
🔷 Fae handles employee requests. People ask Fae in the OpenFrame portal. It sorts out the simple stuff and turns the rest into a ticket, with the context already gathered.
🔷 One place for the whole job. Device management, patching, software deployment, remote access, Microsoft 365 and Google Workspace admin, compliance and ticketing. Apps for Mac, Windows, iOS and Android.
Who it’s for
Small in-house IT teams, from the solo IT person to a team of five, who want to clear the backlog without hiring. MSPs use it too: one place to manage every client, with the same AI agents.
Pricing
$1 per device per month, with a 14-day free trial.
🔗 Try it free at flamingo.run/openframe
What’s the first IT task you’d hand to an agent? Tell us below, we’re in the comments all day.
Can clients easily compare the latest version against previous feedback to see which changes have actually been implemented?
Congrats on the launch! How does patching work for laptops that are offline? Curious whether the update waits for the next maintenance window when they come back online.
@marianna_tymchuk
Thanks so much! Great question.
It doesn't wait for a separate maintenance window. If a laptop is offline when an update or a scheduled run fires, OpenFrame holds it and runs it as soon as the laptop comes back online, within a reconnect window the admin sets (7 days by default).
What happens next depends on the policy the admin sets for apps that are in use:
- Force Update: the app is closed 5 minutes later and updated.
- Wait: the update holds until the user closes the app. You can add a deadline (1 to 30 days), after which the app is closed and updated. The deadline is counted per device from its first try, so a laptop that was offline doesn't lose its window.
Congrats on the launch. The approval design is the part I'd actually judge this on — I run agents all day, and the review queue is where I lose trust fastest, so the four verdicts with the reason shown sounds like the right way to make approvals quick to scan. The per-activity line makes sense too. Curious how it holds up at scale in practice: when Mingo wants to patch the same thing across a fleet, does that come through as one batched review, or one review per device?
@jcbuilds
Thanks JC, totally agree, the review queue is where trust is won or lost. It comes through as one batched review. When Mingo proposes the same action across many devices, you get a single card listing every device it will touch, with the details for each one expandable, and one Approve or Reject for the whole batch. Once it runs, you see the result per device, including any that failed and why, so a few failures don't get buried in a fleet-wide success. The approval gate itself is fully configurable too. Admins decide per activity what needs a review, and for actions they already trust they can remove the gate altogether so Mingo just runs them.
having an agent handle laptop updates and everyday IT requests could take a lot off a small team's plate. congrats on the launch!
@spotch_founder
Thanks Nick! That’s exactly why we built it. The IT person keeps every laptop, login and Wi-Fi network running, and the only time anyone notices is when something breaks. They’re the backbone of everything digital in a company, and they deserve better than drowning in updates and “quick questions.”
OpenFrame gives them AI that does the grunt work while they stay in charge, and finally some credit for everything they do.
@spotch_founder Thanks so much! That’s exactly the idea: let Mingo handle the routine stuff so small teams can focus on the work that actually needs them.
i run a bunch of agents every day and approval fatigue is what gets me. how does mingo decide what's risky enough to wait, and can an admin tune that line per action?
@tsubasatakeuhi
Thanks Tsubasa, approval fatigue is exactly what we designed around.
Every command Mingo wants to run on a device goes through a review first. It identifies what the command actually does (killing a process, deleting files, uploading data off the device and so on) and checks it against the admin's guardrail policy. The result is one of four verdicts: run it, ask the end user, ask the technician, or deny it. Each verdict shows the reason, so you can see exactly which rule or policy made the call.
And yes, the admin sets that line per activity. Routine, low-risk work can run on its own, while anything destructive or sensitive waits for approval or is blocked. Hard guardrails for things like irreversible destruction or data exfiltration can also override a permissive policy, so the riskiest actions still get stopped.
@michaelassraf the four verdicts with the reason shown is a nice touch, makes it way easier to trust. thanks for the detailed answer!
Making an admin approve every fix is the right default for patching, since one bad update across a fleet costs more than a week of manual work. The catch is approval fatigue once the queue holds dozens of routine patches a day. Do you plan to let admins pre-approve classes of changes, like security patches from a known vendor, while keeping config changes on manual approval?
@qingyun_wu
Thanks for the question! Yes, it's driven by policies, not by approving each fix.
For third-party apps that's live today: you set an auto-update policy per app, and new versions from package managers install on those devices by themselves. The policy also says what happens if the app is open: wait for the user until a deadline, or close it after a countdown.
OS patching is what we're building right now, and it's policy-based too. A patch policy says how security and quality updates install, and an update is put on hold automatically if Microsoft reports a known issue. You can still approve, hold or reject something by hand, but that's the exception, not the daily job.