D-ID unveils interactive AI avatars with real-time conversations

Most AI avatars still suffer from the same problem: a floating head mechanically flapping its lips while the rest of the body stays frozen in place.

AI video platform D-ID just released two new avatar models, Express and Premium+, designed to make digital humans feel far less robotic by adding body movement and real-time interaction.

Moving hands and real-time talks

The two models target different enterprise needs based on training time and visual fidelity.

The Express model is built for speed. It requires only about one minute of training video and replicates a user’s natural head movements while speaking. It is lightweight and aimed at fast video generation.

Premium+ is the company’s flagship model. It takes at least a few minutes of video footage to train, but it can reproduce an avatar with full hand gestures and torso movements. More importantly, D-ID says Premium+ is capable of real-time interaction, opening the door for live webinars and instant translations.

D-ID CEO Gil Perry noted that the company believes face-to-face interaction is the most natural way to talk with AI models, envisioning business professionals creating safe digital twins of themselves.

A full push into enterprise marketing

Alongside the avatars, D-ID launched an enterprise marketing suite with integrations aimed directly at corporate workflows.

The suite includes interactive brand avatars, AI influencers, and video translation support across 30 languages. It also integrates directly with Canva and PowerPoint, alongside an API connection for CRMs and marketing automation platforms.

According to D-ID, personalized video campaigns powered by avatars can boost click-through rates by 30% and lift conversion rates by 35%.

For now, these new models are restricted to D-ID’s desktop studio tools, though the company plans to roll the technology out to its mobile app eventually.

Why it matters

For sales and marketing teams, this lowers the bar for scalable outreach. Instead of booking a studio and recording dozens of tailored pitches, a single executive can clone themselves in minutes and generate localized, gesturing video messages in 30 languages with a CRM trigger.

From an industry perspective, D-ID had to make this move. Competitors like HeyGen have been capturing enterprise mindshare with expressive body movements and fluid lip-syncing. D-ID, which initially made waves with single-photo animation, needed full-body realism to stay relevant in lucrative B2B contracts.

We think embedding directly into PowerPoint and Canva is a smart distribution play. Marketers rarely want to adopt entirely new software for every task, but adding a talking avatar directly to an existing presentation deck reduces friction instantly.

The main question mark is security. As real-time avatars with realistic hands and torsos become accessible via standard marketing APIs, verifying identity and preventing executive impersonation will become an increasingly urgent headache.