Meet the new Anam

If you've visited anam.ai recently, things look different. New identity, new visual system, a proper pass on everything - built together with the team at Reboot. Same technology underneath. If anything, the brand finally caught up to what we're actually building.

Where we started

For the past five years, AI brands have been performing the same reassurance. Hand-drawn characters. Warm gradients. Rounded corners. The subtext: don't worry, this isn't dangerous, it's friendly.

For a lot of products, that logic holds. Consumer AI still needs to earn trust, and warmth does real work there. But for a company like ours, where the technical capability is the point, the reassurance aesthetic starts to work against you. It promises something it can't deliver and obscures the thing you're best at.

The wrong versions

Early in the process, the wrong versions were easier to find than the right one.

There was a version that leaned so hard into the technical it alienated everyone. Matrix green wire meshes and Westworld aesthetics that were completely cold. This communicated infrastructure but forgot to make the user feel trust.

There was a version that was polished and hollow. So consumer-friendly it could have been an airline app. It pretended the technology was simpler than it actually is.

Both failed for different reasons. One had no human in the room whereas the other had no technology in the room.

What we shipped

Anam's interactive avatars, powered by Cara, deliver the most realistic live AI interactions on the market. The brand needed to carry that capability without pretending the avatars were human. We landed on what we've been calling the care label - a visual language borrowed from industrial design with technical specificity and functional honesty at its core. The kind of design you find on care labels, receipts and instruction manuals: objects that exist to specify what something is and how to use it.

One piece of the visual system we're especially proud of: a custom halftone shader we built for presenting our avatars and research. It renders photorealistic faces through a dotted, almost printed treatment - referencing low-level, primitive technology. We open-sourced it as an npm package if you want to play with it.

The same logic extends to how we explain the platform itself on our blogs. Instead of polished marketing illustrations, we use abstract technical diagrams: dashed borders, node graphs, version labels. They communicate infrastructure the way infrastructure actually looks precise, connected, versioned.

It's a way of saying: these are intentionally AI. We're building the best interactive avatar models on the market, and the brand now reflects that directly.

Since launching, the new site is converting 15% of visitors to sign-ups.

What the care label actually means

The visual language is the easier part. The harder part is whether it extends into how the product describes itself.

The industry default is to measure performance on P1: the best case, the number that flatters you. An honest measure is P95. A care label brand publishes the P95. It shows the chain of thought when the agent is reasoning. It puts "I'm an AI" in the interface rather than hoping the user forgets.

We think the developers and enterprises building on our platform deserve to know what they're actually building on. Not an apology for being AI or a performance of being human but instead, an accurate description of what this is and how to use it.

As a design brief, that turned out to be the hardest one to get right.

What's next

None of this changes what Anam does for you today. But the identity finally feels like us closer to the work, closer to the ambition, and much more aligned with where we're taking things. Time to bring it into our lab and build out our design system.