Javed Khan is aware of the codec enterprise from the within. Earlier than becoming a member of Neat as CEO in March, he ran Cisco Collaboration as SVP and Common Supervisor, overseeing a multi-billion-dollar portfolio that included its room machine line. Neat’s co-founder and CTO, Ivar Johnsrud, was one of many engineers behind the Tandberg C90, later the Cisco C90, nonetheless broadly cited because the excessive level of codec design.
Neither is arguing these methods had been poorly constructed. Khan’s argument, set out in a weblog submit this week, is that the mannequin behind them has hit a ceiling. The business’s response has been to scale the issue reasonably than resolve it: extra compute, extra cabling, extra sensors across the similar central structure. Prices and complexity develop. What the room can really do doesn’t.
“The very best model of a legacy mannequin remains to be the previous mannequin,” he writes. “The world has modified, and the potential of AI has shifted the goalposts.”
Why Distributed Structure Works Higher in Massive Assembly Rooms
Khan’s proposed different attracts on his time at Aptiv, the place he ran Clever Techniques and constructed distributed edge compute platforms for automotive, aerospace, and robotics. The comparability he reaches for is a self-driving automobile: cameras, sensors, and compute distributed throughout the car, processing in actual time, co-ordinating with out routing every thing by way of a single central unit.
Massive convention rooms, he argues, have the identical structural drawback. Individuals transfer. Conversations shift throughout an area. A centralised system was not designed to trace that dynamism. Including extra {hardware} round it doesn’t change that. “A centralized system can’t sustain,” he writes. “It wasn’t designed to.”
Neat’s reply is to distribute compute, sensing, and AI inference throughout each machine within the room. Cameras, microphones, and companion models every course of domestically and co-ordinate over IP. Rooms scale by including or eradicating units reasonably than changing core {hardware}. Khan frames this as evolution reasonably than rejection: “This isn’t abandoning the codec. It’s redefining what it turns into.”
He additionally distinguishes Neat’s method from partial makes an attempt by legacy distributors. Some distribute units however nonetheless route processing again to a central unit. That, he argues, simply recreates the bottleneck in a unique kind.
How Edge AI Adjustments What a Assembly Room Can Do
For Khan, AI is a part of the structure, not a characteristic on prime of it. When inference runs on the machine, latency drops to milliseconds, information stays within the room, and efficiency doesn’t depend upon community circumstances. The course he factors towards is rooms that transcend capturing conferences to understanding them: monitoring who’s talking, the place consideration is shifting, how a dialog develops throughout a big area.
He indicators a 2026 product roadmap with out detailing it: “I can’t wait to share our roadmap for the remainder of the yr and among the superb new capabilities we’ve got within the pipeline.”
Can Neat Tackle Cisco and Poly within the Massive Room Market?
Neat enters this argument from a powerful place. The corporate reported $250 million in income forward of Khan’s arrival, gained the 2026 Google Cloud Companion of the Yr Award for Google Workspace: Innovation in April, and helps Google Meet, Microsoft Groups, and Zoom natively. It has additionally joined the Microsoft System Ecosystem Platform, aimed toward bettering safety and manageability for enterprise workspace units.
However Cisco, Poly, and others have deep incumbency in massive rooms, and enterprise patrons transfer slowly. When Khan’s appointment was introduced, Irwin Lazar, President and Principal Analyst at Metrigy, informed UC Immediately the rent places Neat “on a a lot greater stage once they promote to bigger corporations,” whereas noting Khan “actually has his work lower out for him in a extremely aggressive market.” Lazar additionally flagged the central rigidity in Khan’s argument: “there’s additionally the dynamic of how a lot will stay within the {hardware} versus the software program.”
That query will get answered by what Neat ships. The roadmap promised for later this yr is the actual check of whether or not this structure has moved past a place into one thing enterprises can deploy at scale.
