Meet Agora’s Conversational AI Engine: it turns speech into smart replies in ~650 ms—faster than you can say “hello, world.” . Low latency, no awkward pauses—voice-first finally feels first-class!

Moderation isn’t binary (true/false). What’s offensive in one culture may be acceptable in another. What feels safe to one user may feel threatening to another. “There’s no one-size-fits-all for content moderation.”. Google’s Ninny Wan reflects on the global challenge of

What do you want to know?.

In a world racing to deploy AI at scale, Google’s Content Safety team is asking:. “Are we doing this responsibly?”. Google’s NinnyWan explains how Gemini goes through the same rigorous AI safety testing as all prior models — and why that matters.

In the age of AI, what is responsibility? What is safety?. Google’s Content Safety team is tackling these big AI questions. NinnyWan, Product Lead for AI Content Safety, joins the Convo AI World podcast to talk moderation, multilingual models & human-AI collaboration. 🔗👇

RT @NewYork_778: BREAKING: I just tested Agora's @AgoraIO Conversational AI Engine head-to-head with traditional voice AI setups. The diff….

The Convo AI World Podcast is live! We're chatting with Amogha Srirangarajan, CEO of Carbon Origins

The Convo AI World Podcast is live! We're chatting with Andrew Seagraves, VP of Research at Deepgram

RT @AgoraIO: Agora Conversational AI SDK v1.7 is here!. This release pushes real-time agent UX to the next level with AI avatars, voiceprin….

RT @riyazmd774: BREAKING: I just tested a voice AI that replies in 650ms. Not seconds. Not minutes. Milliseconds. @AgoraIO ’s Conversatio….

RT @eyishazyer: Voice AI is everywhere now. But most of it sucks. Awkward pauses, robotic responses, connection issues that make you want….

AI avatars, sharper hearing, and visual memory—all now possible with Agora’s real-time voice SDK. v1.7 isn’t just new features—it’s a leap toward human-level presence in AI agents. Your agents don’t just talk anymore. They listen better, look alive, and remember what they see.

🛠️ Updates across the board, from REST APIs to the toolkit. → REST: New `avatar.enable`, `avatar.vendor`, `avatar.params` fields.→ Toolkit: iOS, Android, Web now support `ImageMessage`, `onMessageReceiptUpdated`, and more. If you're building immersive, multimodal AI apps—this.

🖼️👀👤 Picture messaging for agents (Beta). Trying to get your agent to see what the user sees?. In this release you can allow agents to “see” images that users send. It's simple, send an image URL via the new picture messaging API—and the agent will factor it into future LLM.

👤👀🎧 Selective attention locking is h̶e̶a̶r̶ here!. v1.7 introduces voiceprint recognition to lock onto a single speaker—while suppressing background chatter. This sharper ASR accuracy means agents won't get distracted by side conversations if users are in noisy rooms. It’s a

👤 AI avatars are now supported!. Give your agent a face to go along with its voice. Create agents that look and sound alive—perfect for virtual assistants, in-game characters, and branded reps. v1.7 adds support for animated avatars from industry leading vendors—syncing in

Agora Conversational AI SDK v1.7 is here!. This release pushes real-time agent UX to the next level with AI avatars, voiceprint-based selective attention, and image inputs for multimodal LLMs. 🧵. Let's zoom-in and see why it matters for voice ai devs 👇

RT @TheAIColony: BREAKING: Voice AI just got 3x faster. @AgoraIO's Conversational AI Engine delivers speech-to-response in ~650ms—faster….

RT @AIwithGhotai: BREAKING: The voice AI game just changed forever. Meet @Agoraio's Conversational AI Engine, the platform turning speech….

🔗

Can AI avatars give everyone a personal expert?. @trulience's Head of Growth joins the Convo AI World podcast to unpack:.• Training your own avatar agent.• Why emotional nuance matters.• Where AI avatars are actually useful.• What’s next for human-AI interaction. Full