Plume 2.0: From Wi-Fi Optimizer to Agentic Operating Layer for ISPs

This analysis is the result of multiple executive interviews conducted between July and September 2026: Alan Coleman, Chief Product Officer, Plume (July 16); Metin Taskin, CEO, Airties (August 5); and Chris Griffiths, Chief Technology Officer, Plume, joined by Nicolas Fortineau of Plume's marketing team (September 16). It also draws on Plume's Plume Console announcement of October 13, 2026, and its February 2026 platform update.

Why Plume 2.0, and why now

In under two years, Plume has changed its CEO, CTO, and CPO, rebuilt its architecture, and redefined what it sells. On October 13, 2026, at Network X in Vienna, it launched Plume Console, a workspace where ISP teams build and direct AI agents across Plume WiFi, Security and Care.

The short version: Plume is moving from cloud-managed mesh Wi-Fi pioneer to an AI-native operating layer for the subscriber experience. To get there, it has walked back the bet that defined its first decade, that owning OpenSync was the moat. The moat is now the data and the agents that act on it.

Plume 1.0: won deployment, lost the standard

Plume launched the first cloud-managed Wi-Fi platform for ISPs in 2016 and open-sourced its device middleware as OpenSync in 2018 under BSD-3. The logic was simple: give away the edge, own the cloud. It worked as distribution, landing at Comcast, Charter, Bell Canada, Liberty Global, Deutsche Telekom, Vodafone, and others, and reaching a $2.6 billion valuation in 2021.

As an open source community, it never took off. OpenSync was governed, hosted and trademarked by Plume, rivals had no reason to contribute, and operators rallied around bodies they control: RDK-B, prpl and the Broadband Forum's USP/TR-369. By late 2024 founder Fahri Diner had left the CEO role, Comcast veteran Dan Herscovici took over, and Plume stopped making its own pods to become a pure software company.

The reason is structural. Competing Wi-Fi vendors would have been donating engineering to a rival's sales funnel. ODMs and silicon vendors integrated OpenSync because customers asked, but integration is not contribution. And operators cared about the cloud they were paying for, not the agent's source code. A permissive license maximized adoption, not participation, and those turned out to be very different goals.

The OpenSync reset

The new leadership is candid about the past. Griffiths, CTO since June 2025 after Comcast, Nominum and Cujo AI, told me OpenSync was open source but "really not an open platform": every CPE needed deep SDK and chipset work, and integrating one device could take months. Coleman called the closed, OpenSync-only approach "the wrong strategic decision."

The rebuild touches every layer:

• OpenSync moves to long-term support and becomes one OS option, "OpenSync OS," built on OpenWrt.
• A containerized Plume agent runs on RDK-B and prplOS.
• USP/TR-369 replaces OVSDB, with one data model across all three.
•  The monolithic cloud controller becomes stateless microservices, shipping features in days.

Plume device and cloud stack, before and after:

Source: Maravedis 

Every layer on the right is either an industry standard or built to run on one; Plume's differentiation now sits in the cloud services and agents on top.

Plume joined RDK as a Preferred Plus Member in September 2025 and also joined the prpl Foundation. This is the least flashy and most important part of Plume 2.0: adopting USP removes the old "open source, single vendor" objection. Griffiths says operators that once rejected Plume over OpenSync are now taking a fresh look.

The practical payoff, according to Griffiths, is speed and choice. Integration times should drop to a fraction of today's, operators get the same features whether they run RDK-B, prplOS, OpenSync OS or a mix, and ISPs regain freedom to pick silicon and ODMs. That matters more than ever with memory prices squeezing gateway bills of materials. Plume already certifies dozens of devices each quarter with more than 30 ODM partners, prioritized by what its ISP customers want to deploy. Coleman adds that Plume intends to contribute some of the best capabilities it built inside OpenSync to both RDK and prpl, so other operators can benefit too.

What Plume 2.0 sells

Coleman's test for every roadmap item is whether it "turns up on the P&L of ISPs": lower cost to serve, higher ARPU, less churn. Griffiths describes three pillars, with a fourth coming next year.

Wi-Fi, measured by outcomes. Plume is moving from quality of experience to quality of outcome (QoO) and adding L4S, to close the gap between gigabit service and apps that still freeze.

Practical features count as much as algorithms. Griffiths highlighted extender placement guidance: subscribers can walk their home with the app, and technicians can size a home on day one, with the platform continuing to flag devices that sit too far from a node.

Security and device intelligence. A data science team of 10 to 12 builds device fingerprinting so the platform can steer devices correctly and apply policy, such as isolating IoT devices. Plume Security Labs' "SuperProxy" research on streaming boxes enrolling homes in proxy networks now protects every ISP on the platform.

Plume says its security blocks six million threats a day from the gateway, with nothing to install on devices, and out-of-home protection that extends parental controls and threat blocking to the cellular network is on the roadmap. Coleman went as far as calling Plume's security product equivalent to or better than Cujo's. That is a vendor claim worth testing in RFPs, but it shows how seriously Plume now competes in security.

Plume Care. The Sweepr acquisition, led by Coleman as Sweepr's CEO, is the hinge. Coleman calls it a "stacked defense" on cost:

1. Fix the issue automatically before the subscriber notices.
2. Proactively tell the subscriber what is being fixed.
3.  Resolve it through guided self-service.
4. Hand a human agent full context and a diagnosis.

He is refreshingly unhyped: "I'm not trying to AI wash Plume." LLMs handle conversation only, while fixes stay on proven, deterministic paths. Plume Care also runs on non-Plume telemetry, including deployments on rival Wi-Fi platforms, and TELUS has prevented more than 550,000 calls with it. Operators' top ask, Coleman says, is ARPU growth, led by backup connectivity and security.

The economics explain the approach. Deterministic self-service journeys resolve problems reliably, but typically cover only 35% to 40% of inbound volume, because most subscribers still pick up the phone, and traditional IVR resolves under 10% of what it handles. Conversational AI tethered to proven fixes is how Plume plans to capture the rest. The February 2026 platform update added bring-your-own LLM support, natural-language authoring of support flows and an "LLM Fabric" built on the Model Context Protocol.

Plume is also pitching existing Plume Care customers on consolidating their multi-vendor stacks into one platform. On the ARPU side, Plume worked with Charter on its Invincible Wi-Fi backup offer and is building similar options, including access points that double as fixed wireless backhaul, for other ISPs.

Plume Console: an agentic workforce for the ISP

In our interviews, Coleman and Griffiths described an upcoming "Operator Suite" that would replace the separate Haystack tools with one pane of glass per persona. It shipped as Plume Console. It sits on data from more than 50 million homes, roughly 600 million devices and 3,000 profiled applications across 450+ ISPs.

Teams can deploy Plume's pre-built agents, create their own with an agent builder, or modify them with SQL and templates. Agents can flag, notify or resolve issues, and connect to ticketing, CRM and digital care through webhooks, MCP and APIs. Every issue an agent resolves can become an automation, so the share the platform fixes on its own should grow.

Coleman frames the AI in two modes. The first is governed, prescribed workflows that keep Wi-Fi optimized and security always on, resolving most issues consistently. The second, where the Console is aimed, is open-ended investigation of problems with no established fix, with agents investigating and recommending while people set the guardrails and make the decisions. Teams can validate that an agent gives reliable guidance before letting it act. Griffiths adds a collective-defense angle: what the platform learns on one network protects every network, and the Console shows teams what was found and what was done about it.

Plume cites 31% lower churn, 20% fewer technician visits and 16% higher NPS in homes using its technology, with NOS as launch partner. The methodology is not disclosed, so operators should ask how the comparison groups were chosen. Still, this is the right product: it finally puts Plume's data in the hands of people who work it daily, and gives Tier 2 and Tier 3 ISPs tools only the largest operators could build themselves.

Griffiths gave a product-planning example I found telling: because Plume fingerprints devices accurately, an ISP can watch how fast a new flagship phone is adopted on its network and decide where to accelerate Wi-Fi 7 gateway rollouts. Large Tier 1s with their own tooling can skip the Console UI entirely and consume the same documented APIs.

What I heard behind the press releases

The ISP's P&L is the product spec, and Plume knows its limits. Coleman admits Plume historically did not do enough to prove that its subscribers churn less and tolerate price increases better. He runs customer councils for smaller ISPs and is constantly traveling to Tier 1s. He was also clear about boundaries: Plume will not move into the WAN or rebuild tools like ThousandEyes; it will ingest their telemetry instead.

Consolidation is the sales motion. Griffiths argues an ISP would otherwise need one vendor for mesh, another for security and a separate care platform, then glue them together. Coleman adds that agents work better on one shared data model.

Privacy is becoming architecture. Coleman's long-term vision is more intelligence on the router, where observations leave the home but raw telemetry does not. For now, operators want less gateway memory, not more.

Plume acts as a data controller on behalf of its ISP customers and says it complies with GDPR and relevant ISO standards. Agentic AI on rich home telemetry will still draw regulatory scrutiny, especially in Europe.

The doors are reopening. Griffiths described "a lot of tailwinds," including talks with operators that previously rejected Plume because they did not want OpenSync, and recent wins such as Mobinet in APAC. If that holds, the architecture reset should show up as new logos within a year or two.

Uprise, the quiet MDU play. Uprise lets an ISP run Plume's mesh, security and device ID across a building and manage it as one property, with Harmony coordinating channels and power between units. Griffiths cited MSP Internet Subway, two large North American customers rolling out, and two in APAC. It is deliberately not a Ruckus or Meraki competitor. In the U.S. multifamily market, Plume will need to tell this story louder and name customers to be taken seriously.

Deployment mirrors the home model: the ISP specifies a device, often a ceiling-mounted unit per apartment, with an ODM, Plume's software runs on it, and Hub gives the ISP a property-level management interface. Uprise is a "good enough plus" option for ISPs already serving buildings in their footprint, and its security, device intelligence, and care layers are real differentiators in multifamily, where support costs run high. MSP partners I speak with value the platform but want Plume to adapt it faster to their needs.

The competitive picture

The bigger long-term threat is that operator stacks and chipset SDKs absorb enough management that a separate intelligence layer looks optional. Plume's answer, join them and compete above them, is the right one.

Two other fronts are worth noting. Retail-born mesh brands like eero bring strong consumer recognition into ISP deals, which Plume counters with an ISP-branded experience and operator control over data and support. And horizontal AI customer-service platforms can automate simple care interactions cheaply. Plume's argument is that they cannot reboot a router, change a Wi-Fi channel or read home telemetry inside an auditable framework an ISP can stand behind.

Open questions and risks

• Execution. Integrating Sweepr, migrating to a new cloud and protocol, and launching agents at once, while keeping OpenSync on long-term support.
• Lock-in has moved, not vanished. The edge is open now; the value sits in Plume's data, models, and agents. How portable are an ISP's agents and data?
• Agent governance. Agents that reboot 100,000 gateways need audit trails, staged rollouts and rollback.
• Proof. Churn and NPS gains need operator-published results, not just vendor studies.
• Runway. A 2021 valuation of $2.6 billion still has to be grown into.

Two more belong on the list. Privacy rules in Europe will shape how far agentic AI on home telemetry can go, and Coleman's on-device vision depends on gateway memory costs that currently point the wrong way. And Uprise needs named customers before U.S. multifamily MSPs and property owners will treat it as a serious option.

The Maravedis take

Plume 2.0 pairs humility about standards with ambition about AI, and that is the right combination. OpenSync succeeded as a deployment vehicle and fell short as a standard; Plume has accepted that its real asset was never the code on the gateway but the data and the know-how to act on it.

The strategy looks strongest when set against Plume's closest rival. When I spoke with Airties CEO Metin Taskin in August, he described the home Wi-Fi management market as "mostly a two-horse race" between Airties and Plume. The two are now converging on the same destination from opposite directions.

Airties took the standards road first, rebuilding its management layer on EasyMesh, TR-181 and USP years ago, and Taskin argues Plume is now following that path. On the device layer, he has a point. Plume went up the stack first instead, building a closed loop from telemetry to care and buying Sweepr, a company Airties had partnered with rather than acquired.

The result is an asymmetry that favors Plume for now. Plume has closed most of the standards gap; Airties has not yet closed the care and engagement gap. Airties remains the safer choice for Tier 1s that want to stay firmly in control of their stack and their care operations, and its pending Aprecomm acquisition gives it a clearer path into lower-ARPU markets.

Agentic AI on its own is no longer a differentiator: Airties launched its Aura agentic engine five weeks before Plume Console. The contest will be decided on data quality, agent governance, the cost of running AI for every home, and who owns the care conversation. On that last point, Plume leads.

Verdict

For the next 12 to 24 months, Plume 2.0 has the stronger proposition for operators that want a vendor to own outcomes, especially Tier 2 and Tier 3 ISPs and any operator whose priority is cutting care costs or selling security. The condition is execution: Plume must complete its architecture migration and prove its outcome numbers with operator-published results.

Plume 1.0 asked operators to adopt its stack. Plume 2.0 asks them to adopt its intelligence, on whatever stack they already run. That is a much better bet.

Sources

• Interview with Alan Coleman, Chief Product Officer, Plume, July 16, 2026.
•  Interview with Metin Taskin, CEO, Airties, August 5, 2026.
• Interview with Chris Griffiths, Chief Technology Officer, Plume, with Nicolas Fortineau, Plume marketing, September 16, 2026.
• Plume, New Plume Console Multiplies ISP Business Results by Putting AI Agents to Work on Cost to Serve, Churn Reduction and Revenue Growth, October 13, 2026.
• Plume, Plume to Deliver the Only Open Agentic AI Platform that Supports the Complete ISP Customer Journey, February 26, 2026.
• Plume, Plume joins RDK community as Preferred Plus Member.
• Plume, Plume joins prpl Foundation.
• Plume, Plume acquires Sweepr.
• Plume, SuperProxy research.
• Tommy Flanagan, Rethink Research (Faultline), "Plume acquires Sweepr for WiFi support: A missed break for Airties?"
•  Advanced Television, Airties introduces Aura: Agentic AI Engine, September 8, 2026.