Flamingo, a brand new MSP vendor, made its OpenFrame platform generally available this week after a lengthy beta period.
Flamingo is taking on the likes of ConnectWise, NinjaOne, Kaseya, HaloPSA and N-able with a platformn that is a mix of open-source AI and internally developed tools.
“The idea behind Flamingo is to create much more efficient MSPs,” Flamingo founder and CEO Michael Assraf said. “We cover the vendor part with open source, which MSPs can self-host or get from us on a managed basis at a pretty low price. We have about eight modules in the product, including RMM, MDM, and remote access control. We charge a fraction of what they’d pay buying these as separate tools. All those tools are orchestrated in a way that’s very easy for AI agents to consume. The agents are already integrated with the full suite of actions and data.”
Assraf, who also founded cybersecurity vendor Vicarius in 2016, has raised $6.7 million for Flamingo including a $4.5 million seed round led by Vertex Ventures in September to go with a $2.2 million pre-seed round led by Focal a year ago. OpenFrame has been in beta since last May. Assraf said more than 400 MSPs used the beta to manage more than 10,000 end points.
Flamingo’s greatest challenge may be convincing MSPs to switch from established tools vendors they have used for years.
“Taking on such a large portion of the MSP stack makes it harder, because we’re not an AI agent bolted on top of existing tools,” Assraf said. “But our architecture and design are very modern from an AI perspective, which makes it easier for us to build agents that really work. Everyone says ‘you just click a button and everything happens.’ On our platform it actually works that way, because of how the AI can access actions and data.”
OpenFrame is available as a managed SaaS service or a free self-hosted version. MSPs choosing the SaaS service deploy the agents to end points with no servers to maintain. The self-hosted version is full open-source tooling installed with a guided CLI public Terraform modules for AWS and Google Cloud, and it runs without the AI agents. The self-hosted version is free. MSPs that want the agents can move to the paid tier.
OpenFrame uses two AI agents: Mingo and Fae. Mingo acts as a technician to handle scripts, policies, queries and incident response. Fae handles end-user requests from a desktop app on each device, and each MSP can rename it for its own clients.
MSPs on the agent tier can choose to run OpenFrame on OpenAI or Anthropic models. They buy OpenFrame tokens and can spend them on their frontier model of choice. Monthly plans at $1 per device include 10 million tokens. Annual plans at $0.80 per device include 25 million tokens, with additional tokens costing $10 per million. Costs of typical tasks range from an estimated 2 cents for answering a “how do I” question to 57 cents to diagnose a slow or failing device.
“What we see is that once MSPs adopt our AI at scale, they don’t look back,” Assraf said. “They don’t even compare us to other tools. When we’re compared on feature parity, that’s harder for us, because those other tools have been around for years. But AI with more granular access to actions and data is much stronger and more efficient at handling problems.”
In a case study posted on the Flamingo web site, beta tester Brandon Clegg of Clegg Technologies said he anticipates OpenFrame helping his MSP win clients.
“Just giving that presentation of how it works, how it functions, is the biggest piece,” Clegg said. “It'll actually at least give us the upper hand against most other companies when we're doing bids, simply because knowing that AI is the direction that the world is essentially moving towards in IT. I think it would be ideal for a lot of my clients and a lot of future clients to be able to make that decision to move with us versus somebody else.”