Platform Update: More Natural. More Deployable. More Scalable.
By Michelle Collins, Chief Operating Officer, CodeBaby
Listen to any real conversation and you’ll notice how little of it follows a script. People interrupt each other, trail off, circle back, switch topics mid-sentence, and sometimes switch languages entirely. (And if you’re like me, sometimes we jump in a little too eagerly in the pauses.) We toss in little “mm-hmms” that aren’t really interruptions at all, just signals that we’re listening. None of this feels remarkable to us because it’s simply how humans talk.
Technology should support that kind of interaction, not force people to flatten their communication to fit the machine.
That belief has guided our latest platform update. Rather than chasing features for their own sake, this release focuses on three things: making conversations with our avatars feel more natural for the people having them, making avatars easier to deploy in the physical and digital spaces where organizations actually work, and making the whole platform simpler for developers and partners to build on.
Conversations That Feel More Human
The most noticeable change in this release is how naturally our avatars can now handle the messiness of real conversation.
Start with language. Visitors no longer need to select a language before they begin talking. Our avatars can now be set to recognize the language someone is speaking and continue the conversation in it, whether that’s English, Spanish, Portuguese, French, Japanese, or one of dozens of other supported languages. We’ve also expanded our speech recognition with OpenAI’s Whisper, which brings our avatars to communities that conversational AI too often overlooks.
Interruptions were the other big focus. Users can now cut in while an avatar is speaking, ask a follow-up, or steer the conversation somewhere new without waiting for a scripted response to wind down. Just as importantly, the platform can tell the difference between a genuine interruption and a casual acknowledgment or a bit of background noise, so the conversation keeps its rhythm instead of stopping and restarting at every “uh-huh.”
Individually, these might sound like small refinements. Together, they close a lot of the distance between talking to software and simply talking.
Built for the Real World
When we started this work, avatars mostly lived on websites. That’s no longer the world we’re building for. Our customers are now deploying conversational experiences in hospitals and classrooms, hotel lobbies and retail floors, trade show booths, kiosks, and even holographic displays.
This release meets them there. CodeBaby avatars now support Proto holographic displays and other portrait-oriented devices without requiring a supporting website, which makes it much easier to create immersive experiences in physical spaces. And through new APIs, webhooks, and expanded data capabilities, avatars can now pull in and present dynamic information in real time, whether that’s real estate listings, product details, or live customer support data.
The through line in all of this is that a deployment should feel purposeful in its environment, not like an experiment someone wedged into the corner of a lobby.
Built for Developers and Technology Partners
Making conversational AI easier to build matters just as much as making it easier to use, and this update includes a lot for the people doing the building.
A new public Avatar API, expanded webhook capabilities, support for Markdown knowledge files, and backend infrastructure improvements all shorten the path from idea to working integration. We’ve introduced a usage-based pricing option for customers who want more flexibility in how they consume the platform, and security enhancements aligned with SOC 2 best practices give enterprise teams additional confidence as they deploy.
If you’re embedding conversational AI into an existing application, or building something entirely new around it, our goal is for the platform to feel like a partner in that work rather than another system to wrestle with.
Progress Toward a Bigger Vision
Every release is a snapshot of what we believe conversational AI should be. This one reflects a conviction we’ve held from the beginning: that this technology should feel natural rather than mechanical, that it should fit into products without adding complexity, and that it should scale to genuinely meaningful interactions with customers, employees, students, and patients around the world.
We’re not all the way there yet. No one in this industry is. But watching an avatar smoothly follow someone from English into Japanese, or hold a conversation that bends and flows the way human conversations actually do, it’s hard not to feel like the gap is closing.
We can’t wait to see what our customers and partners build with it.
