LaunchedEditorial Listing

OpenFrame

Flamingo · OpenFrame: Flamingo's AI Agents for MSP and IT Operations, with RMM, Patching and Phone Approvals

Open OpenFrame

OpenFrame, from Flamingo, is an IT operations platform for MSPs and small IT teams with two built-in AI agents: Mingo runs fleet-wide jobs for technicians, such as patching, scripts, device queries and log-driven fixes, and Fae helps each end user from a desktop app on their own computer. Both act directly on OpenFrame's own RMM, remote access, patching, security policy and ticketing tools, and anything your rules mark ask first waits for a technician's approval in the mobile app. It costs $1 per managed device per month, with unlimited technicians.

PricingPaid
Setupmedium
Runs onWeb · Desktop
APIYes
Open sourceNo
DocsYes
CategoryProductivity
MSPRMMEndpoint ManagementPatch ManagementRemote AccessIT Service ManagementHelp DeskTicketingHuman-in-the-LoopMobile AppMicrosoft 365Google WorkspaceSource-AvailableSmall Teams

Best for

Small and mid-size MSPs, MSSPs and lean in-house IT teams managing Mac, Windows and Linux fleets who want to replace several per-technician tools with one per-device platform and let AI agents do routine fixes under approval rules they control

Not ideal for

Organizations that need SOC 2 attestation or non-US data hosting today, MSPs that depend on built-in PSA client billing, backup or email security, and teams that want to self-host the AI agents rather than only the underlying tools

Who it's for

Owners and technicians at small and mid-size MSPs and MSSPs, and lean in-house IT teams managing employee computers

Capabilities

  • Mingo, the technician agent, takes plain-language requests across the whole fleet: device and software queries, scripts, security policies, patching and incident response, and Flamingo says it watches device, cloud and help desk logs around the clock to fix problems before a ticket is filed
  • Mingo writes, tests and saves scripts for the script library, can test edited scripts with saved secret variables instead of credentials pasted into chat, and runs scripts on one device or many, now or on a schedule
  • Fae, the end-user agent, runs as a desktop app in the menu bar or system tray and handles slow machines, approved software installs, printer trouble, account unlocks and how-to questions, gathering logs and system info and handing off to a technician when it should
  • MSPs can rename and customize Fae per MSP and per client, so each client's staff see the MSP's own helper
  • Per-action, per-client rules set what runs on its own and what is marked ask first; commands aimed at more than one device always need approval, and a run is rejected if any target device is set to deny that command
  • Approvals arrive in the iOS and Android apps with the command shown as a readable block that Mingo can explain line by line, and every action, approver and timestamp is written to an audit log
  • Remote control of Mac, Windows and Linux computers built on MeshCentral, from the desktop app or a phone
  • Software deployment and patching through Chocolatey, Homebrew and WinGet, with per-device wait-and-notify or forced-countdown update behavior
  • Security rules for disk encryption, firewall, admin rights and MFA built on Fleet MDM, with drift put back automatically, plus osquery-based device queries
  • Ticketing, logs and alerts (OpenFrame SIEM with Grafana Loki) and client documentation and runbooks in the same platform, all reading and writing one shared device record
  • Microsoft 365 and Google Workspace tenant connections that sync users, devices, licenses and apps into one client view, with identity resolution and offboarding workflows (added in v1.6.10 in November 2026; Flamingo's homepage still lists user and license sync as a Q4 2026 item)
  • Web console, macOS and Windows desktop apps, and iOS and Android apps for checking the fleet and approving actions on the go

Limitations

  • Mingo and Fae run only on the paid managed service; the free self-hosted edition is the tooling layer without the AI agents
  • The code is source-available under the Flamingo AI Unified License, not an OSI open-source license: forks must stay under the same license, published changes may be reused and relicensed by Flamingo, and offering the software as a competing managed service needs a commercial license
  • Young product: general availability came in July 2026 and paid billing in late October 2026, and Flamingo labels its Gen 2 and Gen 3 tool sets (SaaS management, cloud backup, IAM, email security, zero trust access, then backup and disaster recovery, EDR and network monitoring) as coming soon
  • PSA billing for invoicing your own clients is still on the roadmap, so MSPs need to keep a separate billing tool for now
  • SOC 2 is not complete; Flamingo's trust center tracks the audit as in progress
  • The hosted service runs on Google Cloud in the US only, with no other data region documented
  • AI usage beyond the monthly token allowance is billed pay as you go, so heavy agent use adds cost on top of the device price

Use cases

  • Asking Mingo which computers are not encrypted across every client and turning encryption on, with the change waiting for a technician's approval
  • Scheduling an overnight update of one app on every Mac and approving the multi-device command from a phone
  • Letting employees ask Fae to fix a printer, unlock an account or install an approved app on their own computer instead of calling the help desk
  • Having Mingo write, test and save a remediation script, then run it on a schedule across a client's devices
  • Consolidating RMM, remote access, patching and ticketing for a small MSP onto one per-device bill
  • Troubleshooting a client device with Mingo while the client is still on the phone

Our take

OpenFrame's case rests on a structural point: its agents can do real work because they sit on the same RMM, patching, policy and ticketing record a technician uses, rather than calling out to tools they do not control. Combined with a flat $1 per device and unlimited technicians, that is an aggressive offer for a small MSP paying several per-seat vendors. The risk is maturity. The platform went GA in mid-2026, important MSP pieces like PSA billing and backup are still roadmap items, there is no SOC 2 report yet, and hosting is US-only, so it is easier to justify for small MSPs and internal IT teams than for clients with compliance requirements. The public code helps with transparency, but read the license before treating it as an exit path, and note that self-hosting leaves the agents behind. The sensible way in is the one Flamingo's own customer stories describe: run it on a slice of endpoints next to your current RMM, start with strict approval rules, and loosen them as you watch what Mingo and Fae do.

Who should use it

Owners and technicians at small MSPs and MSSPs, and lean in-house IT teams, who manage Mac, Windows and Linux computers, are tired of paying several per-technician tools, and want AI agents to handle routine fixes, patching and user requests under approval rules they set.

Who should skip it

MSPs serving clients that require SOC 2 or non-US data hosting today, teams that rely on built-in PSA billing, backup or email security, larger enterprises looking for a full ITSM with change and asset management, and anyone who needs to self-host the AI agents.

Strengths

  • Agents act on the platform's own RMM, remote access, patching, policy and ticketing tools, so a request can run end to end instead of producing advice
  • Per-action, per-client approval rules, mandatory approval for multi-device commands and an audit log of every action
  • Simple pricing: $1 per device per month with unlimited technicians, both agents and an AI token allowance included
  • 14-day trial with no card and no contract
  • Fae can be rebranded per MSP and per client
  • Source code is public, and the tooling layer can be self-hosted for free

Weaknesses

  • The AI agents are only available on the hosted service
  • Source-available license with copyleft, contribution-back and no-competing-service terms, not OSI open source
  • Young platform: GA in July 2026, with PSA billing, backup and email security still on the roadmap
  • No SOC 2 report yet, and US-only hosting
  • AI usage past the included tokens is billed on top of the device price

OpenFrame pricing

Monthly

$1 per device

Billed monthly

  • Billed on your average device count each month
  • Unlimited technicians; Mingo and Fae included
  • 10M AI tokens a month included, then pay as you go
  • No contract, cancel anytime

Yearly

$0.80 per device per month

Billed annual

  • Device package paid yearly; devices above it at $1 each
  • Unlimited technicians; Mingo and Fae included
  • 25M AI tokens a month included

Free tier limits: 14-day free trial of the managed service with no card and no contract. The self-hosted edition of the tooling layer is free but does not include the Mingo and Fae agents.

Note: Extra AI usage is pay as you go at $10 per million OpenFrame tokens ($100 buys 10M), with optional auto top-up and a configurable spend limit; each AI model converts tokens at its own rate, and Flamingo says simple lookups and queries use no tokens. The $1 price covers today's Gen 1 tool set (RMM, remote access, patching, PSA ticketing, SIEM, MDM, documentation and automation). Flamingo's pricing page lists Gen 2 at $3 per device (SaaS management, cloud backup, IAM, secure email gateway, zero trust access, marked coming in Q4 2026) and Gen 3 at $5 per device (adding backup and disaster recovery, EDR, endpoint protection, XDR, MDR and network monitoring, marked Q1 2027), both as coming soon.

Where OpenFrame excels

Fleet-wide encryption check

An IT admin asks Mingo which computers are not encrypted; Mingo queries every device through the shared record, proposes turning encryption on and saving recovery keys, and the change waits for approval on the admin's phone.

End-user self-service through Fae

A client's employee asks the MSP-branded Fae app for an approved app or a printer fix on their own laptop; Fae gathers logs and system info, makes the fix under the MSP's rules, and opens a ticket for a technician when it cannot.

Gradual migration off an existing RMM

An MSP installs the OpenFrame agent on one client's endpoints next to its current RMM, compares results during the 14-day trial, and moves clients over one at a time.

OpenFrame vs. competitors

OpenFrame vs. Serval

Serval is an AI-native IT service management platform for mid-size and large companies: its help desk agent resolves employee IT and HR requests in Slack, Teams or email by running workflows admins have published, and it fixes devices through connected MDM and identity tools. Pricing is sales-led after a pilot. OpenFrame is aimed at MSPs and small IT teams, brings its own RMM, remote access, patching and policy tooling, and is priced per device with a self-serve trial.

OpenFrame vs. Nexus EDGE

Nexus EDGE pairs Yellow.ai's cloud agent with a desktop client that runs fixes on the employee's own computer after the employee approves each change, which is close to what Fae does. It is an enterprise product sold through demos and connects to existing IT systems such as ServiceNow and Okta. OpenFrame adds a technician-side agent for fleet-wide work and replaces the RMM and patching stack itself, with public per-device pricing.

Frequently asked questions

What is OpenFrame, and how is it related to Flamingo?

Flamingo is the company and OpenFrame is its product: an IT and security operations platform that combines device management, patching, remote access, ticketing, logs and security policies with two AI agents, Mingo for technicians and Fae for the people they support. It runs in the browser, as a desktop app for macOS and Windows, and on iOS and Android. Flamingo also runs OpenMSP, a community for MSPs.

How much does OpenFrame cost?

$1 per managed device per month on monthly billing, based on your average device count, with 10 million AI tokens a month included. Annual billing is $0.80 per device per month for a prepaid device package, with 25 million tokens a month, and devices above the package are $1 each. Technicians are unlimited and both agents are included. Extra AI usage is pay as you go at $10 per million tokens, with an optional auto top-up and spend limit. Every team starts with a 14-day free trial with no card and no contract.

Which actions do Mingo and Fae need approval for?

You set a rule per action and per client: let it run, or ask a technician first. Anything marked ask first, and any command aimed at more than one device, waits for approval in the OpenFrame iOS or Android app, where the command is shown as a readable block before you tap approve. If any device in the group is set to deny that command, the whole run is rejected. Every action, approver and timestamp is logged. Flamingo's pages do not describe this identically: the FAQ and product page say routine actions you allow can run on their own, while the trust center says both agents always ask before changing a device. Enforcement is also recent; v1.6.0 (October 30, 2026) fixed a bug in which Mingo-issued commands bypassed the policy check that requires approval. Start with strict rules during the trial and confirm what actually runs unattended.

Is OpenFrame open source?

Its code is public on GitHub, but under the Flamingo AI Unified License rather than an OSI-approved open-source license. You may use, modify, self-host and redistribute it, but forks and network deployments must stay under the same license, Flamingo may reuse and relicense modifications you publish, and offering the software as a paid hosted or managed service requires a separate commercial license. Bundled upstream projects such as Fleet, osquery and MeshCentral keep their own licenses.

Can OpenFrame be self-hosted?

The tooling layer can. Flamingo provides a CLI that creates a Kubernetes cluster (k3d locally, or GKE or EKS) and deploys the platform with Argo CD, and the self-hosted version is free. Mingo and Fae, however, run only on the paid managed service, which is hosted on Google Cloud in the US with an isolated environment per customer.

Which AI models does OpenFrame use?

OpenFrame is model-agnostic: the product page says teams pick models from OpenAI or Anthropic, and the trust center lists Anthropic, Google Gemini and OpenAI as subprocessors for underlying language models. AI usage is metered in OpenFrame tokens that can be spent on any supported model, with each model converting tokens at its own rate. Flamingo says it does not use customer data to train models and that each AI interaction is scoped to the customer's own environment.

Is OpenFrame SOC 2 compliant?

Not yet. Flamingo's trust center tracks its SOC 2 audit through Vanta as in progress and not yet complete, with NIST CSF 2.0 and CMMC 2.0 planned. Gated security documents are shared under NDA on request.

Integrations & fit

Microsoft 365Google WorkspaceMeshCentralFleetosqueryChocolateyHomebrewWinGetGrafana LokiAPI
Good fit forSolo / individual, Startup / small team
Pricing modelPaid· Paid subscription required
See pricing on OpenFrame →

Alternatives to consider

About OpenFrame

OpenFrame is Flamingo's bet that an AI agent is only useful in IT if it can touch the same systems a technician does. Rather than a chat window added to someone else's RMM, OpenFrame bundles the tools an MSP usually buys from several vendors (device inventory and scripting, remote access built on MeshCentral, patching through Chocolatey, Homebrew and WinGet, Fleet-based security policies, ticketing, logs and client documentation) around one shared device record, and puts two agents on top. Mingo works for technicians: you ask in plain words for a fleet-wide job, such as finding unencrypted laptops or updating Zoom on every Mac tonight, and it queries devices, writes and runs scripts, patches and watches logs for problems. Fae lives in the menu bar or system tray of each managed computer and handles the user's own requests, handing anything it cannot fix to a technician; MSPs can rename it for each client. Control comes from per-action, per-client rules: an action is either allowed to run or marked ask first, and commands aimed at several devices always wait for a tap in the iOS or Android app, where the command is shown before it runs. The price is the other draw: $1 per device per month with unlimited technicians, both agents and an AI token allowance included, after a 14-day trial with no card. The tradeoffs are real. OpenFrame reached general availability in July 2026, PSA billing, backup and email security are still on the roadmap, the SOC 2 audit is unfinished and the hosted service runs only in the US. The code is public under Flamingo's own source-available license, and the free self-hosted edition does not include Mingo or Fae.

Updates from OpenFrame

LaunchOpenFrame launches on Product Hunt

Flamingo launched OpenFrame on Product Hunt, presenting the Mingo and Fae agents, phone approvals for risky changes and pricing of $1 per device.

New Featurev1.7.0: Mingo-first interface redesign

OpenFrame 1.7.0 replaced the single sidebar with a three-column layout that keeps Mingo conversations visible while technicians work on devices, tickets or lists, rolled out to every SaaS tenant, along with fixes for mobile sign-outs and Fleet agent installs.

New Featurev1.6.10: Microsoft 365 and Google Workspace sync, semantic guardrails and AI spend limits

This release synced users, devices, licenses and apps from Microsoft 365 and Google Workspace into one client view with identity resolution and offboarding workflows, moved AI command guardrails from regex matching to a semantic policy engine, and added configurable auto top-up thresholds and 30-day spend limits for AI tokens.

PricingPaid billing goes live

OpenFrame v1.6.0 launched billing for all customers and made Mingo the starting point of the dashboard. Flamingo says beta accounts moved to the same 14-day free trial as new teams on November 4, 2026, followed by $1 per device per month or $0.80 on annual billing.

Milestone$4.5M seed round led by Vertex Ventures

Flamingo announced a $4.5 million seed round led by Vertex Ventures, bringing total funding to $6.7 million after a $2.2 million pre-seed led by Focal in October 2025. Flamingo says the seed funds its first go-to-market team and the next two generations of OpenFrame.

LaunchFirst general availability release

OpenFrame v1.0.4 was the first general availability release: access codes were removed from sign-up, Flamingo's native RMM agent replaced TacticalRMM, and Fae gained per-MSP and per-client customization.

Are you the founder? Claim this listing →