Agora

Congratulations to the OpenAI team on the launch of GPT-Realtime-2! To celebrate the launch, @masonthemaker from the Agora team put together a few demos powered by Agora’s Conversational AI Engine showing what low-latency voice AI experiences can feel like in practice.

How @AgoraIO is simplifying the Voice AI future for you

@AgoraIO at AI institute

@DeepgramAI @Microsoft @GroqInc @AssemblyAI @GoogleAI @AWS Which voice AI stack should you test first? Agora’s Conversational AI Benchmark gives developers a live view of end-to-end stack performance across latency, cost, and speed × cost tradeoffs. Use it to compare options before you build. agora.io/en/conversatio…

@DeepgramAI @Microsoft @GroqInc 💰𝗟𝗼𝘄𝗲𝘀𝘁 𝗣𝗿𝗶𝗰𝗲 | 𝟭,𝟴𝟱𝟭.𝟭𝟱 𝗺𝘀 | $𝟬.𝟬𝟭𝟭/𝗺𝗶𝗻 - ASR: @AssemblyAI Universal 2 → 514 ms - LLM: @GoogleAI Gemini 2.0 Flash → 1224.1 ms - TTS: @AWS Polly (Standard) → 113.05 mms

Developers keep asking: “Which model should I use for my voice agent?” The better question is: “Which full voice AI stack performs best end to end?” Agora’s Conversational AI Benchmark ranks real-world STT + LLM + TTS stacks by latency, cost, and overall tradeoff.

Convo AI World Podcast: Faraz Siddiqi Co-Founder, CTO @ Bluejay x.com/i/broadcasts/1…

@AgoraIO x @MurfAIStudio: Real-time voice agents, redefined. We’re partnering with @MurfAIStudio to bring expressive, low-latency TTS into the Agora Conversational AI Engine: ⚡ Ultra-low latency 🌍 Multilingual at scale 🗣️ Natural, human-like voice 🔄 Real-time, interruptible conversations Built on SDRTN®—making voice AI instant, reliable, and production-ready. Learn more: tinyurl.com/3eev6seb #VoiceAI #ConversationalAI #TTS

Voice is the default interface for next-gen AI🎙️ During #AnimeJapan Week, @AgoraIO teams up with @MiniMax_AI, @AkoolInc, Sentino, @VCUBE_official & @OceanBaseDB for Voice AI Agent Builder Night in Tokyo on🗓Mar 30 | 18:00–21:00 JST 🔹 Learn how startups build Voice AI 🔹 Try LLM-powered AI devices (R1-4G) 🔹 Meet Tokyo’s AI builder community From anime-inspired characters to real-world AI companions—this is where ideas come to life. 👉 Register here: luma.com/ac213tp3 #voiceAI #developer #convoAI #Japan

📣 𝗩𝗼𝗶𝗰𝗲 𝗔𝗜 𝗠𝗶𝘅𝗲𝗿 𝗡𝗬𝗖! @AgoraIO & @DeepgramAI are hosting a Voice AI Mixer 📍 NYC (details upon RSVP) 📅 Tuesday, March 24 🕠 5:30–8:30 PM 𝗔𝗴𝗲𝗻𝗱𝗮: Live Voice AI demos No-pressure networking Drinks & bites Register now!👇 #VoiceAI #NYCMeetup #AIMixer

.@seeedstudio @pollenrobotics @nvidia held a hackathon post GTC this weekend. Our clear favorite was this robot programmed to act as the dungeon master for your DnD game.

We made Jarvis, a robot assistant built for the 2026 Seeed Embodied AI Hackathon in sc. Huge thanks to the amazing sponsors and ecosystem partners that made this possible: @seeedstudio @NVIDIA @PollenRobotics @huggingface @AgoraIO! 1/3

Had such a fun time hacking on reachy mini and making friends along the way! Thanks to @seeedstudio @nvidia @huggingface @pollenrobotics @AgoraIO for organizing

@NVIDIAGTC week starts today. The MiniMax team will be around the Bay Area all week, meeting builders, researchers, and founders. If you are around, come find us: • Mar 16-18 afternoons: GTC Booth #142 w/@gmi_cloud • Monday (Mar 16) 7pm MiniMax x @baseten AI Leader Dinner • Wednesday (Mar 18) 6pm Exclusive night w/@novita_labs • Wednesday (Mar 18) 6pm OSS AI @ GTC • Wednesday (Mar 18) 6:30pm Founder Salon w/@gmi_cloud • Thursday (Mar 19) 9:30am Physical AI Meetup w/@AgoraIO • Friday (Mar 20) 6:30pm Agentic AI Leaders Dinner • Saturday (Mar 21) 2pm MiniMax AI Founder Day in SF✨ If you're building with frontier models, multimodal AI, or real-time agents — let's talk! 👉Follow MiniMax event calendar for real-time updates: luma.com/MiniMax

This week in Voice AI may have been quieter, but it still brought some noteworthy updates 🤗 ! @AgoraIO has released its new ConvoAI Studio providing a no-code deployment tool to configure, test, & deploy Voice AI agents, along with targeted solutions for Customer Support and Sales & Marketing. Additionally, a new open-source solution called Dograh has emerged, which mimics VAPI or Retell AI solutions. @togethercompute has also enhanced its offerings by adding hosted for speech-to-text (STT) and text-to-speech (TTS) models to deploy very low latency agents in a similar way to Cloudflare approach. Two opensource TTS models coming from @hume_ai and @FishAudio promising zero content hallucinations with a novel text-audio sync approach in the first case and promising high level of control and emotions in the second case. As usual, upgrades for all the major orchestration frameworks @livekit, @pipecat_ai and @TenFramework and found specially interesting the approach to split instructions by modality added to LiveKit that we have been using at @livetok_ai for a while too, and the interesting blog post from Daily showing the performance of the new @nvidia Nemotron model for Voice AI use cases. Full update with more news, details and link below 👇 Have a nice week!

If you are attending NVIDIA GTC, check this meetup too Register here: luma.com/zdrsjwd8 Wednesday 18 March 17:30 - 20:30 GMT-7 Location: 2804 Mission College Blvd Santa Clara, California #NVIDIAGTC #VoiceAI #AIMeetup @AgoraIO

@aivalleyio’s first Demo Day!! 🩵👏 Im partnership with @UofBeta @AgoraIO @theagentic @VictorSuOrtiz @zifeng_liang

Agora Partners with Sentino to Advance Physical AI Through Customizable, Retentive AI Agent Experiences 𝐊𝐧𝐨𝐰 𝐌𝐨𝐫𝐞 👇 smartsolutionsworld.com/agora-partners… @AgoraIO #Agora #SmartSolutionsWorld