It’s been a few years since the world has had access to ChatGPT, Gemini, Claude and other advanced AI tools, and those concepts are now becoming a key feature on the CEDIA Expo show floor as manufacturers release both integrator and user AI features designed to make installation and usability as easy as possible.
AI-generated speaker and lighting designs, AI-assisted proposals, natural language programming for setup and scene creation, and of course home control were all front and center as the industry embraces AI head on.
AI popped up everywhere: on the show floor, on the Smart Stage, in CEDIA Expo programming, and more. The technology has the ability to revolutionize the way integrators work and how homeowners interact with their homes, so it’s impossible to ignore. Here’s a recap of what stood out from conversations with Crestron, Kaleidescape, Savant, Josh.ai, Sonos, WAC/Modern Forms, and D-Tools/Portal.io.
Crestron: AI as an Integrator’s Junior Designer
Crestron has built AI directly into Configure Pro, its system design and programming tool. The pitch is straightforward: feed it a natural-language prompt, or even a sales proposal or an Excel equipment list, and it will automatically build out rooms and devices in the software. The model is trained on years of Crestron’s own data, plus input from internal teams and beta testers.
Crucially, Crestron built in a guardrail that came up more than once across the show: the AI won’t make changes without asking first. It proposes; the human approves. And for now, this tool is squarely aimed at integrators and dealers, not homeowners. The goal is speed and profitability on the business side of the industry, not a new consumer-facing assistant.
Savant: AI That Programs the Whole House
Savant had one of the more ambitious AI stories at the show. Their platform now includes a genuine AI personal assistant woven into the smart home system, controlling lighting, HVAC, entertainment, shades, thermostats, scenes, energy, and air quality through plain conversation. It runs on touch panels, push-to-talk handheld remotes (deliberately avoiding always-listening microphones), and the companion app, and it supports more than 56 languages.
The headline feature, though, is what they’ve done with Blueprint, their design tool. Upload an architect’s floor plan and electrical/lighting drawings, and the AI reads the document, identifies rooms, windows, and fixtures, and matches architect-specified products to real Savant and GE SKUs. In one demo, it mapped out several rooms and dozens of fixtures, then auto-drew and fully programmed the entire house in a single click, compressing what would normally take weeks into minutes. From there, the design flows straight into a one-click e-commerce quote with a complete bill of materials, and Savant frames this as a way to standardize quality across dealers rather than leaving best practices up to individual installers.
Josh.ai: A Decade-Old AI Bet, Now Baked Into Everything
Josh.ai leaned hard into the idea that it’s been doing this longer than most. The company’s pitch at its gallery presentation was that Josh has spent more than ten years building it into actual client homes through AIXOS, the platform that ties together its app, touchscreens, remotes, and voice control into one system.
The most tangible AI feature is natural-language automation. Instead of manually building out a scene, an integrator (or eventually a homeowner) can type or say what they want, like turning off the lights at 8 p.m. on weekends and kicking on the fireplace if it drops below 50 degrees, and Josh will parse that into triggers, actions, and conditions automatically, all of which stay fully editable afterward. In one live demo, the team asked Josh to build a football-watching scene (blue lighting, pop music, the right game on) purely from a rambling verbal description, and the system pulled it together in seconds.
Josh also showed off a “room aware” version of this on its Edge remote line, where a button can be programmed just by talking to the system: “Hey Josh, program button six to turn the AC down to 60 degrees.” The remote itself is designed to be swappable between rooms and use cases, and Josh says battery life runs up to 60 days.
On the integrator side, the company’s Pro Studio dashboard now includes an AI-powered help center. Rather than digging through documentation to figure out how to integrate a specific TV brand or reset a remote, dealers can type a plain-language question and get a direct answer pulled from Josh’s own support library. It’s a smaller feature than the automation work, but it points to the same theme running through a lot of this year’s AI announcements: using AI to cut down on the busywork of running an integration business, not just to add a flashier voice assistant.
Sonos: Multi-Assistant AI
Sonos came to CEDIA Expo with its broadest announcement slate in some time, and AI sits at the center of it. The company unveiled Sonos 27, a platform-wide software update, alongside new hardware: the Beam Ultra soundbar, Sonos Ace Ultra headphones, a new line of multi-functional portable speakers, and Amp Multi, a professional multizone streaming amplifier aimed squarely at the custom integration channel.
Rather than building one proprietary voice assistant, Sonos is opening the platform up to multiple AI providers at once. Sonos 27 introduces two new control layers: Sonos 27voice, an in-house assistant focused on music and mood, and Sonos 27mcp, a framework that lets outside AI agents plug directly into a user’s system. OpenAI is the first outside partner under this model, letting ChatGPT users play music on Sonos hardware without leaving the ChatGPT interface. Sonos also previewed Sonos Custom Agents, a forthcoming feature letting users build their own assistants, each with its own model, persona, and voice, to control the system. For integrators, this signals that Sonos wants voice control treated as a swappable layer rather than a single locked-in stack, which could change how systems get specified and supported down the line.
WAC / Modern Forms: AI for Visualizing Lighting Before You Buy It
WAC’s contribution is Inspire, an in-house AI design tool built for lighting visualization. Designers and integrators upload a photo of an actual space, a client’s home gym, say, and then prompt the AI to add or modify decorative fixtures directly on top of the image: “add more tiers,” “make it bigger,” “put this into a home gym.” The tool responds to iterative, conversational prompting, and the more detail it’s given, the more refined the output becomes.
The target audience is integrators and interior designers who want to stay inside WAC’s catalog while mocking up looks for clients quickly. Once a design is finalized, it flows directly into WAC’s ordering pipeline. Designers can send the AI-generated concept straight to WAC’s team for a quote.
D-Tools + Portal.io: AI Built on Combined Industry Data
The other big AI storyline heading into the show didn’t come from the floor itself but from news that broke just ahead of it: D-Tools and Portal.io, two of the integration industry’s most widely used software platforms, are merging. D-Tools CEO G Paul Hess stays on to lead the combined company, while Portal.io founder Kirk Chisholm becomes chief product officer over the full portfolio. Both companies plan to introduce their joint leadership team to CEDIA Expo attendees for the first time.
Portal.io already markets itself as first to market with an AI proposal builder, and Chisholm says that tool is already getting reworked as part of the merger. But both executives were pointed about not wanting to chase AI for AI’s sake. Chisholm described resisting the temptation to treat AI as something you “bolt onto a feature” just to market that you have it, saying the combined company intends to be “way more thoughtful than that.”
What actually has Chisholm excited is the data the merger unlocks. Individually, D-Tools and Portal.io each have years of aggregated historical project data; combined, that dataset becomes something neither company could fully exploit alone. The vision is AI that works quietly in the background rather than announcing itself, letting an integrator ask, through whatever interface they’re already using, to have a proposal built or a drawing generated based on documented best practices drawn from that combined data.
Looking further out, Chisholm predicts the integrator’s software stack will trend toward an “agentic” model within a few years, describing a future where dealers effectively hire digital agents to handle fractional or even full-time workloads inside the business, rather than just using software as a tool.
Kaleidescape: AI as a Hardware Economics Story
Not every AI conversation on the floor was about software features. Kaleidescape’s take was about market forces: the AI boom is a major reason solid-state storage is so expensive right now, since massive capital is pouring into AI infrastructure and competing for the same manufacturing capacity. Their read on the market is that this won’t last. As storage capacity expands and some of that AI investment cools off, they expect SSD prices to start easing, possibly by late 2027 or 2028.
It’s a useful reminder that AI’s influence on this industry isn’t limited to products. It’s also reshaping component costs and supply chains behind the scenes.
AI on the Smart Stage
Worth a quick mention: AI also got its own dedicated programming on the Smart Stage this year, not just booth demos.
One panel, “When Smart Homes Get Smarter: AI Enters the Rack,” looked at how AI is working its way into AV, security, and control platforms. It was co-sponsored by Josh.ai and Savant and featured Josh.ai CEO Alex Capecelatro alongside Savant EVP Craig Spinner and a pair of integrators. This conversation touched on what is currently capable with AI in the home and what will inevitably make its way to your customers’ homes and demands.
A second panel focused on the business side, covering how integration companies are already using AI for things like proposals, documentation, service diagnostics, and marketing. This of course featured Ari Supran of Sonance and a trio of other integration firms that are using AI to help them do their jobs.
We’ll have more on these sessions later, along with the full audio recordings.
What It All Means
A few things stood out. The clearest one: AI is mostly being used to save integrators time, not to impress homeowners, turning a proposal or a floor plan into a finished design so dealers can move faster and take on more work.
There’s also a real split on how AI should show up for the end user. Some companies want to own that experience with their own built-in assistant. Others are opening things up to whatever AI a customer already uses, rather than building one proprietary voice. Either way, the bet is that people will eventually just talk to their homes.
And one merger in the space made the case that the real advantage isn’t the AI model itself, it’s the data behind it. Years of combined project history is what actually makes an AI proposal or drawing useful.
Put it together and AI at CEDIA Expo this year looked less like a gimmick and more like plumbing, quietly speeding up design, quoting, and programming while the industry figures out who ends up owning the conversation with the homeowner.
