Javed Khan, CEO of Neat, is a technology executive with more than two decades of leadership experience spanning enterprise collaboration, intelligent systems, edge computing, and cybersecurity. He became CEO of Neat in March 2026 after serving as Executive Vice President and President of Intelligent Systems at Aptiv, where he led work across advanced intelligent and edge-based technologies. Before Aptiv, Khan spent more than nine years at Cisco as Senior Vice President and General Manager of Collaboration, overseeing a major period of evolution in enterprise communications and collaboration technology. Earlier in his career, he spent 14 years at Symantec as Vice President of Enterprise Security, giving him a background that combines collaboration platforms, intelligent systems, and enterprise security.
Neat is an Oslo-based technology company that designs intelligent video collaboration devices for modern meeting spaces. Its portfolio combines purpose-built cameras, microphones, displays, audio systems, and processing hardware with an AI-powered distributed architecture designed to make hybrid meetings more natural and easier to manage. Neat’s devices support major collaboration platforms including Microsoft Teams, Zoom, and Google Meet, as well as bring-your-own-device configurations, with AI capabilities used for functions such as intelligent framing, audio optimization, and adaptive meeting experiences. The company focuses on bringing more intelligence directly into the meeting room while offering organizations hardware that can be deployed and managed across spaces ranging from small focus rooms to large conference environments.
After leading major collaboration initiatives at Cisco Systems and working on intelligent edge systems at Aptiv, what made the CEO role at Neat the right next chapter for you, and how does your past experience shape your vision for the company’s AI transformation?
Building products that get used at scale has always been at the heart of what I do, and that’s how you find out fast whether you’re actually solving a problem.
At Cisco, I watched video conferencing go from a boardroom luxury to the backbone of how the world works. At Aptiv, I saw intelligent edge computing quietly transform entire industries — automotive, aerospace, robotics — in ways most people didn’t notice until suddenly they couldn’t imagine working without it. What connected both was the same thing: intelligence moving closer to where things actually happen.
And that’s exactly where we are now. In the last couple of years, AI models have become small enough and powerful enough to run directly on a device sitting in a conference room. That’s not an incremental improvement. That’s a fundamental shift in what’s possible, and frankly, it’s an incredibly exciting time to be in this industry.
That’s why Neat made sense. The founding team started from scratch in 2019 and built an architecture designed natively for the AI era, not layered on top of decade-old legacy code. We have over 600,000 devices deployed across 20,000 customers in 90 countries. We’re becoming the global standard. And there’s still so much more to build.
You’ve described the conference room as one of the most underestimated surfaces in AI today. Why do you believe the physical meeting space is becoming strategically important again after years of software-first collaboration trends?
Coming out of the pandemic, we saw something interesting happen. Conference room adoption accelerated. People came back to the office and the meeting room became central again, not just for video calls, but as a gathering place for teams to think, decide, and work together.
But here’s the thing. Even with that renewed relevance, these devices still only get used while you’re in a meeting. You walk in, join a call, walk out. The room goes dark. And that’s a massive missed opportunity.
What’s changed is that we can now do so much more. Now that powerful compute resides in the room—along with AI capabilities that are simple, well-designed, reliable, and open—the space is useful in all the moments that used to fall through the cracks. Before the meeting starts. After it ends. When a local team just needs a space to think out loud without a meeting link in sight.
The conference room has always been a hub for collaboration. AI is what finally lets it act like one.
During your time leading Webex through the pandemic-era remote work boom, what lessons did you learn about how people actually collaborate versus how software platforms assumed they would collaborate?
The pandemic forced the industry to move fast, and collaboration platforms did a remarkable job improving the experience for people working from home. Audio got better, video got better, the interfaces got simpler. Real progress was made for the person sitting at their kitchen table.
But here’s what happened at the same time. The gap for the person in the conference room got wider, not narrower. While the home experience improved significantly, the room experience presented a different set of challenges that were harder to solve from the software layer alone. Remote participants were consistently disadvantaged in ways the platform couldn’t see and didn’t measure: missing the side conversation before the meeting started, unable to read the room, left out when decisions drifted off-camera.
That’s exactly why we’ve invested so heavily in features like Intelligent Layouts and Intelligent Framing at Neat. Intelligent Layouts dynamically adjusts how remote participants are displayed in the room, giving everyone equal presence regardless of where they’re sitting. Intelligent Framing continuously reads the room and adjusts the camera view automatically, so the remote participant always sees the right person at the right moment without anyone in the room touching a thing. Our goal is simple: the far-end participant should be just as satisfied with their experience as the people sitting in the room.
Many AI collaboration tools today focus on transcription, summaries, and copilots. What do you think they are still fundamentally missing about real-world meetings?
Transcription and summaries are valuable, and the industry has done an excellent job establishing documentation as a standard. With that baseline in place, the next opportunity is to use AI to actively improve the meeting experience itself.
Think about what actually happens before a meeting even starts. For a lot of people, just joining a meeting can be stressful. Wrong platform, wrong link, unfamiliar interface. We hear this from customers constantly. That friction happens before a single word gets transcribed, and it’s something we’re actively working on with our platform partners to make better.
Consider what happens during the meeting itself. Some people dominate, others disengage, remote participants feel like they’re on the periphery. Post-meeting AI is powerful for documentation and follow-through, but the meeting experience itself also needs attention while it’s happening. That’s a live, in-the-moment problem, and it requires intelligence that lives in the hardware itself.
And then there’s everything after. Action items that sit in a summary nobody reads. Decisions that don’t make it into the right workflows. AI should be connecting those outcomes to the rest of the enterprise automatically, not generating another document for someone to manually process.
That’s what we’re focused on at Neat: the before, the during, and the after. Better experiences powered by AI that is working in the room, not just documenting what happened in it.
How do advances in edge computing and large language models change what conference room hardware is capable of doing locally versus relying entirely on the cloud?
For most of the history of AI in collaboration, the cloud was where intelligence lived, and it remains the backbone of how we manage and coordinate at scale. What’s changed is that we can now do more by bringing some of that intelligence closer to where the meeting actually happens.
The first is speed. You can’t have meaningful latency in audio processing or camera framing. When the AI has to make a round trip to the cloud and back, you feel it. The response has to be instant, and that means the processing has to happen right there in the room.
The second is privacy. When AI runs directly on the device, the most sensitive data never has to leave the room. That’s not just a technical detail; it’s a fundamentally different privacy posture. A policy document can’t give you that. The architecture has to.
The third is cost. Running AI locally on the device reduces dependence on expensive cloud GPU infrastructure. That’s a real saving that compounds across a large device fleet, and it’s something IT leaders are increasingly paying attention to.
Running powerful AI models directly on the hardware, rather than as a remote API call, is now possible in a way it simply wasn’t a few years ago. You get the best of both worlds: the cloud handles management and coordination at scale, and the edge handles the experiences where speed, privacy, and local cost efficiency matter most. That shift is a big part of what pulled me back into this industry, and to Neat.
You’ve worked across industries, including automotive, aerospace, robotics, and enterprise collaboration. Are there concepts from intelligent systems in those industries that you believe will eventually reshape workplace collaboration technology?
The automotive parallel is the one I keep coming back to. Autonomous driving is one of the most sophisticated edge computing solutions ever built. The sensors detect what’s happening in real time, the car makes decisions that genuinely affect people’s lives, and all of it is processed at the edge in the moment. The response has to be instant because the consequences of getting it wrong are immediate.
A meeting room is the same problem in a different environment. The room changes every time. Different number of people, different acoustics, different lighting, different conversation dynamics. The hardware has to read what’s happening and respond gracefully, without anyone having to manage it.
What draws us to edge devices at Neat comes back to three things: speed for the experiences that demand it, privacy because the most sensitive data shouldn’t have to leave the room, and cost because local processing reduces dependence on cloud infrastructure at scale. The discipline of designing for real-world variance rather than ideal conditions is something the automotive and robotics industries have grappled with for decades. Enterprise collaboration is now catching up.
Why do you believe the next major innovation wave in enterprise collaboration may come from hardware and edge intelligence rather than from the software platforms dominating headlines today?
I’d push back on the framing slightly. It’s not hardware versus software, the two have to work together. But what has changed is that hardware is now capable of doing things that simply weren’t possible a few years ago, and that creates a genuinely new opportunity.
The platforms, Microsoft, Zoom, Google, are doing remarkable things inside the meeting. But there are things that hardware enables that software alone cannot fully replicate. Framing the room accurately. Processing audio with the precision that comes from microphones and speakers purpose-built for the space. Handling everything locally without a round trip to the cloud. When intelligence is built natively into the hardware, you get a level of performance and privacy that software alone cannot achieve in the same way.
When you build that intelligence into the hardware, you make the software better. Better framing means the remote participant gets a cleaner, more accurate view of the room. Better audio processing means voices are captured more clearly, which improves everything downstream. Better local processing means faster, more private, more reliable experiences for everyone in the room and everyone joining remotely.
So the question isn’t which layer wins. The question is how do you build a hardware platform that makes the software better, and a software ecosystem that makes the hardware more valuable. That’s the bet we’re making at Neat, and it’s why we’ve built from the ground up rather than trying to retrofit AI onto legacy architecture. The companies that figure out how to make hardware and software work together seamlessly, that’s where the next wave comes from.
As AI agents become more capable, how do you see the role of meetings evolving? Will meetings become more intentional, more automated, or potentially less frequent altogether?
What you’ll see first is a transformation of everything around the meeting and the meeting experience itself. Before the call, agents will help prepare participants, surface relevant context, and ensure the right people are in the room. After the call, agents will act on outcomes automatically, assigning action items, updating workflows, connecting decisions to the rest of the enterprise without anyone having to chase it down. The meeting becomes less of an isolated event and more of a node in a continuous, intelligent workflow.
The second thing agents will change is how IT manages infrastructure. When agents can access and manage room hardware directly, the cost and complexity of running a meeting room estate drops significantly. IT teams spend less time on maintenance and troubleshooting, and more time on strategic priorities. That’s a meaningful shift for any organization running meeting rooms at scale.
What won’t change is the importance of the meeting itself as a space for genuine human collaboration and deliberation. The conversations that require judgment, that need people to read the room and make decisions together, those don’t go away. If anything, as agents handle more of the administrative weight, the meetings that remain become more focused and more valuable.
The room hardware has to be ready for all of this. As agents become active participants rather than just support tools, the device in the room becomes the physical window through which those agents engage with the people in it.
Companies increasingly worry about privacy, governance, and security when AI systems are embedded directly into workplace environments. How should organizations think about balancing intelligent collaboration with trust and control?
This is one of the most important questions in our industry right now, and it’s only going to get more pressing as AI becomes more embedded in the workplace.
The starting point has to be architecture. When you embed AI in a room with persistent sensors and always-on processing, the question of where the data lives is not a detail. If the most sensitive data never leaves the device, you’ve addressed a category of risk that no privacy policy alone can touch.
At Neat, security and privacy are built into how we design and engineer our products from the ground up, not bolted on afterward. Neat OS is a purpose-built operating system engineered specifically for meeting spaces, with hardware-backed secure boot, verified operating system integrity, and strict process controls that limit what software can access even if a vulnerability is discovered elsewhere. All audio and visual AI processing is handled locally on dedicated edge processors within volatile memory, guaranteeing zero data retention, zero cloud transmission, and zero exposure of customer information.
That architecture also simplifies something organizations increasingly care about: compliance. When AI processing stays on the device, data sovereignty becomes far easier to demonstrate. It removes entire categories of risk from the compliance conversation and makes it significantly easier to meet the requirements of frameworks like HIPAA for healthcare, SOC 2 for service organizations, and CCPA and CPRA for organizations operating under California privacy law. The room becomes an asset in your compliance posture, not a liability.
What AI is changing is the speed of security, not the direction of our strategy. The window between discovering a vulnerability and exploiting it is shrinking fast. That’s why we’ve moved beyond periodic reviews toward continuous monitoring, ongoing vulnerability management, and AI-assisted code analysis throughout our development process.
For organizations evaluating technology partners, the most useful question isn’t whether a vendor is attached to the latest AI security initiative. It’s whether security is embedded deeply enough into everything they build to withstand whatever comes next. At Neat, that’s been our focus from day one.
Looking ahead five years, what does the ideal AI-powered conference room actually look and feel like, and what capabilities do you believe will seem obvious in hindsight once this transition fully takes hold?
The thing that will seem most obvious in hindsight is this: no one should ever have a subpar meeting experience simply because they’re joining remotely. That’s the bar we’re building toward, and we’re closer than most people think.
You walk into the room and the meeting is ready for you. No fumbling with cables, no troubleshooting audio, no adjusting the camera. Everyone, whether they’re in the room or joining from home, has equal presence throughout the conversation without anyone having to manage it.
After the meeting, outcomes don’t sit in a summary nobody reads. Action items are assigned, workflows are updated, and decisions connect automatically to the rest of the enterprise. The meeting becomes a node in a continuous, intelligent workflow rather than an isolated event.
And the infrastructure running all of this is managed largely by AI agents, not IT teams chasing down device issues. Rooms are always ready. Fleet management happens in the background. The technology becomes truly invisible.
What will seem obvious in hindsight is that we waited so long to treat the room as an intelligent system rather than a collection of peripherals. The idea that you’d need to troubleshoot audio or manually frame a camera before a meeting could start will look as unnecessary as it actually is.
Thank you for the great interview, readers who wish to learn more should visit Neat.
