Topic hub
26 Vox Roboti reports on Voice AI. Everything the desk has filed, newest first.
Deepgram is moving its voice AI models directly onto disconnected laptops. Cloud hyperscalers hate it. Enterprise security teams finally get speech tech without the surveillance.
Marla Quinn · 5d ago · 5 min read
Anthropic launched Claude Duet, turning its reasoning engine into an autonomous desktop worker. The move takes direct aim at legacy RPA and voice-driven software control.
Devon Achebe · 6d ago · 5 min read
Apple's 27.2 operating system updates give Siri conversation recall, deep file indexing, and five new languages across modern hardware.
Devon Achebe · 6d ago · 6 min read
Meta is reorganizing its future around an AI agent named Muse, pairing the technical pivot with a fresh stock plan to bankroll the staggering compute costs.
Ray Dalton · 10d ago · 5 min read
Meta's new voice agent Muse pulled ten million users in six weeks. Now Zuckerberg plans to monetize the compute with merchant fees, premium tiers, and smart glasses.
Tobias Reinholt · 11d ago · 5 min read
Deepgram moved heavy voice model workloads to AWS Trainium. The move cuts deep into Nvidia's margin buffer and rewires voice agent unit economics.
Marla Quinn · 11d ago · 5 min read
At nineteen, Zach Yadegari turned TikTok arbitrage into a $30M nutrition exit. His new $10M bet, Persona, trades calorie photos for ambient voice agents and custom hardware.
Ray Dalton · 16d ago · 4 min read
McKinsey is making hundreds of undergrad applicants pitch a simulated client for ten minutes. Humans still score the tape, but the standard interview is officially dead.
Cass Iverson · 18d ago · 5 min read
Peter Mo landed $26M from Luminous Ventures and Huaqin to build Ambi, an ambient voice agent running across Apple hardware while dodging two-party consent minefields.
Devon Achebe · 20d ago · 5 min read
Google's Gemini voice migration is breaking hands-free dialing on closed clamshell phones, leaving drivers stuck with a dashboard prompt to flip their handsets open.
Selma Ortiz · 22d ago · 5 min read
Hardware startup Natura unveiled Interface, a $99 smart ring designed to route voice prompts to Claude, ChatGPT, and Grokbot while undercutting Oura on price.
Ray Dalton · 24d ago · 5 min read
A year inside Amazon's Alexa Plus overhaul reveals faster home automation, sharp latency cuts, and a screen experience ruined by aggressive display ads.
Tobias Reinholt · 24d ago · 5 min read
Co-op Legal Services has wired OpenAI into its probate desk to grade calls with grieving families across 50 data points, turning empathy into a tracked metric.
Ray Dalton · 24d ago · 4 min read
An Android APK teardown reveals Google is testing Gemini Calling to outsource routine personal phone calls to an AI agent.
Selma Ortiz · 27d ago · 5 min read
ElevenLabs just dropped its v4 model, scaling synthetic speech to 90 languages and handing developers real dynamic control over human vocal nuances.
Cass Iverson · Sep 28, 2026 · 4 min read
ElevenLabs crossed $600 million in annual recurring revenue while charting a path toward voice AI that mimics raw human emotion.
Cass Iverson · Sep 25, 2026 · 5 min read
Mati Staniszewski is talking margins, IPO patience, and why synthetic callers must identify themselves up front. Here is what the voice giant's candid playbook means for enterprise tech.
Devon Achebe · Sep 24, 2026 · 4 min read
Meta is bringing New York Times Cooking to its smart glasses, ditching chatbot hallucinations for structured voice prompts you can actually use while basting a roast.
Ray Dalton · Sep 24, 2026 · 5 min read
Tencent built Gander with a split anatomy: a fast cerebellum for real-time speech and a swappable brain for deep work. It interrupts less, but the trade-offs bite.
Devon Achebe · Sep 20, 2026 · 5 min read
Qwen launched Qwen3.8-Omni-Flash at $0.15 per million input tokens, undercutting Google Gemini Flash by 80% while matching multimodal benchmarks.
Devon Achebe · Sep 19, 2026 · 4 min read
Apple brought active noise cancellation and live speech translation to the $129 AirPods 5, then botched the lineup with an absurd twenty-dollar upsell.
Cass Iverson · Sep 18, 2026 · 5 min read
Respeecher engineers unpacked the hidden backend failures that wreck production voice agents, from generic REST tools to markdown readouts over phone lines.
Tobias Reinholt · Sep 17, 2026 · 4 min read
Stability AI released Stable Audio Open, skipping pop songs to give sound designers a 47-second synthesis engine trained on lawsuit-free audio.
Ray Dalton · Sep 17, 2026 · 6 min read
Apple unveiled Apple Intelligence at WWDC, overhauling Siri, tethering ChatGPT to iOS 18, and locking the entire system behind a steep hardware upgrade wall.
Ray Dalton · Sep 17, 2026 · 5 min read
OpenAI cut out the middleman in voice AI. If your entire startup was three APIs stitched together with duct tape and a markup, your business model just vanished.
Nina Falco · Sep 17, 2026 · 5 min read
Simulating sympathy on support calls doesn't calm people down. It insults them. Build competence into your speech stack instead of breathy dramatics.
Marla Quinn · Sep 14, 2026 · 5 min read