AI Is No Longer Optional

AI-powered features are now standard in leading video conferencing platforms. For example:

• Zoom Rooms software has introduced Smart Gallery, “which displays multiple video feeds from a single conference room. This allows the Zoom Rooms cameras to focus more closely on groups of participants and display these people more clearly to remote attendees.” The platform also offers AI Companion transcription and voice command options that are constantly being updated and enhanced.
• Microsoft Teams Rooms now offers Cloud IntelliFrame, “a new experience that allows online meeting attendees to see people in Teams Rooms more clearly through the smart video feeds of in-room participants. These smart video feeds are created by zooming into the faces of the in-room participants and by eliminating distractions, thereby enhancing the hybrid meeting experience.”  Many users are likely already familiar with the platform’s speaker recognition capabilities and Microsoft® Copilot® software integrations, too.

These features are cloud-enabled, but many are performance-critical — they require real-time responsiveness (especially for functions such as framing and speaker switching), demand low latency and high reliability, and require local data handling for privacy and compliance.

That’s why spaces need AI that operates at the edge — in the device, in the room.

Where Edge AI Makes the Difference

In fact, there are some projects where edge AI solutions are an absolute must. The reasons become abundantly clear when you consider use cases such as:

• Executive boardrooms: In these installations, you need instant speaker recognition and real-time transcription with zero lag, even in fast-paced discussions. These systems need to handle multiple cameras, mics, and streams simultaneously without “round trips” to the cloud or performance degradation.
• Hybrid classrooms: These projects require live captioning that stays accurate without falling behind.
• Government or finance: Local processing ensures compliance and data sovereignty, in addition to offering offline functionality for secure or air-gapped environments.

There’s another issue with the expanding adoption of these solutions — namely, most existing processors weren’t built for AI workloads. They struggle to run multiple AI models, such as audio, video, and transcription; they can overheat or lag in environments where they’re always “on, and they can deliver inconsistent performance under heavy data loads.

An Answer: The Crestron Collab Compute

After taking all of this into consideration, Crestron developed a new purpose-built solution made for AI at the edge: the Collab Compute with Intel® Core™ Ultra processor. Designed with AI-enhanced collaboration in mind, this powerful “brain in a box” features an Integrated Neural Processing Unit (NPU) that offloads AI tasks from CPUs and GPUs, enabling platform-native AI features without overtaxing the system.

Crestron built the Collab Compute with future-readiness in mind, giving it the power to handle next-gen AI features from Microsoft Teams, Zoom, and more. We made sure that those who invest in our edge-ready hardware can be confident in its compatibility with future platform updates.

When it comes to AI solutions, the cloud is critical — but it’s not enough, especially as user demands evolve and hybrid workers return to the office. As we mentioned elsewhere in the blog post we quoted above:

“This technology is raising people’s expectations for when they go into a collaboration space,” says Hintze. “If I’m used to sitting in my own home office getting meeting transcription notes and all the other benefits of the attendant AI technology, and then I go into the office and meet with people in person, I'm going to expect the benefits of having that agent AI helping and facilitating that meeting.”

Deployments and Connections Made Easy

Beyond the AI issues that the platform helps to solve, there are a multitude of other benefits to Collab Compute.

“When we built Collab Compute, we really built it with the ability to scale,” says Joel Mulpeter, Crestron senior director of product marketing. “The whole goal was to create one core centralized solution and put it in every room type.” To do that, the engineering team needed to create a solution that could “work and play well” with nearly any peripheral that’s brought into the system — while at the same time preserving a consistent experience across even the most far-flung, expanding enterprise. “First and foremost, the user in the room needs that consistency — the same touch interface, same experience and workflow. That consistency of experience has to be preserved if they join a meeting, share content, or do BYOD conferencing.”

Consistency of user experience wasn’t the only desired result, however: consistency of installation was a huge consideration in the development of Collab Compute, too. “We understand that in the current environment, installers and IT departments may be responsible for getting perhaps hundreds of rooms up and running at breakneck speed.” The ability to deploy large numbers of Collab Compute solutions quickly and at scale became a priority.