Hearspeech-to-text15
GoogleModel providermodel release2026-10-04
Google posts DiarizationLM-Gemma-4-E4B-v1 open weights for transcript speaker correction
Half the parameters of the 2024 8B DiarizationLM with lower Fisher WDER (2.99% against 3.28%), small enough for a local post-pass on call transcripts; the card disclaims official Google support, and meeting-corpus gains are under one point.
Details3 sources
Google's Hugging Face organisation published an Apache-2.0 DiarizationLM fine-tune of Gemma 4 E4B that post-processes ASR and diarization output to correct speaker attribution in telephone and meeting transcripts.
DiarizationLM-Gemma-4-E4B-v1
Deepgram lets Nova-3 swap keyterms mid-stream without reconnecting
Multi-intent Nova-3 calls can retarget biasing per step without a reconnect — useful when collections, IVR, or agent handoffs change the active glossary mid-call.
Details2 sources
Deepgram's 2 October 2026 changelog lets a Nova-3 `/v1/listen` session replace its full keyterms list through a Configure message so multi-step calls can change vocabulary without dropping the WebSocket.
Nova-3 mid-stream keyterms
Cactus Compute ships Whistle, a 16.9 MB on-device ASR model
A sub-20 MB CPU ASR that shares Needle's tool-calling binary is a different on-device path from Fermion Phonon-2 (2026-09-29-fermion-phonon-2, 164 MB) — vendor WER claims still need independent checks beyond the blog tables.
Details4 sources · 3 X posts
Whistle is an open speech recognition model that runs on CPU in the same C++ engine as Needle, covering seven languages in a 16.9 MB file with vendor-claimed 11 ms time to first token.
Whistle
Twilio Batch Transcription Configuration reaches general availability
Post-call STT becomes a named, reusable config object with Deepgram Nova-3/Nova-2 or Twilio-managed engines — no redeploy to swap model or language.
Also Connect
Details1 source
Twilio made Batch Transcription Configuration generally available so recorded calls can be transcribed after completion via a reusable config for engine, speech model, language, and destination.
Batch Transcription Configuration
SoundHound Polaris Edge enters the Intel Edge AI Catalog
The listing is a product-availability and distribution milestone for an edge speech component; the catalog describes local endpoint deployment but does not state a fleet-wide customer go-live.
Also Connect
Details2 sources · 1 X post
SoundHound made Polaris Edge available in Intel’s Edge AI Catalog for local voice AI on kiosks, drive-thrus, and embedded devices.
Polaris Edge · vendor Intel
MicrosoftModel providermodel release2026-10-01
Microsoft ships MAI-Transcribe-2-Streaming at $0.54/hr, AA #1
Microsoft’s first dedicated streaming SKU on the Transcribe-2 line posts AA-WER Streaming #1 at $0.54/hr — the realtime counterpart to September’s $0.10/hr batch model.
Details3 sources · 2 X posts
Microsoft AI released MAI-Transcribe-2-Streaming, a low-latency streaming STT sibling to batch MAI-Transcribe-2, covering 60 languages with continuous language detection. Introductory price $0.54 per audio hour through year-end; public preview on Azure Speech / Foundry.
MAI-Transcribe-2-Streaming
Introducing 3 new models: MAI-Transcribe-2-Streaming, MAI-Voice-2.1 and MAI-Voice-2.1-Flash.
Accurate streaming transcription. Natural speech and less waiting between turns.
Build voice agents that keep the conversation moving!
https://x.com/MicrosoftAI/status/2105693024013467905
Microsoft AI has released MAI-Transcribe-2-Streaming, taking the #1 spot for Final Transcript accuracy and First Partial Transcript accuracy on AA-WER Streaming with 2.5% WER at 0.13s after end of speech
MAI-Transcribe-2-Streaming is @MicrosoftAI's new streaming Speech to Text
https://x.com/ArtificialAnlys/status/2105694108736188894
Good Tape raises €600k from Nordic Makers in a buyout from Zetland and launches Explore across transcribed interviews
Finderup says transcription turned into a commodity 18 to 24 months ago, so the €600k funds a memory layer over the archive instead of better recognition; Explore is free at launch and has no revenue test yet.
Also Use
Details2 sources
The Copenhagen transcription service, used by 3 million people, is now independent of the Danish newsroom that built it and adds a free layer that links each new recording to everything a user has transcribed before.
Explore
LokutorModel providermodel release2026-09-30
Lokutor open-sources Oído, open-vocabulary speech recognition on a $5 ESP32-S3
Open-vocabulary STT on a $5 chip replaces fixed command lists, but at roughly twice real time on one core it suits short commands rather than live agents; companion TTS Ito (2026-10-02-lokutor-ito) followed two days later.
Details3 sources · 1 X post
Oído runs NVIDIA's 13M-parameter Conformer-CTC Small for any English sentence, with a Spanish model too, entirely on an ESP32-S3 microcontroller with no cloud or NPU, through a new int8 C inference engine.
Oído
1/ Meet Oído: full speech-to-text running on a $5 microcontroller. Any English sentence, no cloud, no NPU. 3.7% word error rate on LibriSpeech, 2.3× fewer errors than Espressif's own model on the same chip. Open source today 👇
https://x.com/lokutorAI/status/2105260950731587868
KugelAudio Clarity 1 strips background talkers from a call before the voice agent hears them
Background speech that sets off false barge-ins is a common voice-agent failure, and Clarity sits in front of STT rather than inside it. Krisp's VIVA line (2026-05-06-krisp-viva-2, W19) is the nearest filed rival (analytical). The 50 ms cost assumes 240 ms input frames.
Details2 sources · 1 X post
Clarity 1 is a real-time speech enhancement model that removes noise and keeps only the primary speaker, using a short reference recording or else the loudest voice, so speech-to-text and turn detection respond to the caller.
Clarity 1
Devnagri ships Black Bird ASR, claiming lead on Voice of India telephonic bench
Vendor-claimed Indian telephonic ASR lead versus Sarvam and Gemini on an independent closed set — self-run API bake-off, not a third-party leaderboard publish — after the March Speech AI platform launch (2026-03-09-devnagri-speech-ai).
Details3 sources
Devnagri AI announced Black Bird, a new Indian-language ASR model, reporting 8.7 average OI-WER on the Voice of India telephonic benchmark across 15 languages — ahead of Sarvam Saaras V3 and Google Gemini 3 Pro on Devnagri’s own production-API bake-off.
Black Bird
bolipiPlatformlaunch2026-09-30net-new
bolipi ships an offline push-to-talk voice typing beta for Mac
Shipping the models inside a 950 MB download keeps bolipi fully offline, the same privacy pitch as SpeechFast (2026-10-18-speechfast-local-dictation-mac) three weeks later, but it supports English only and names no model, so accuracy claims cannot be compared with Willow or TwinMind's 140-language mode.
Details4 sources
Apple Silicon Mac users hold fn, speak, and get cleaned-up text pasted into whatever app has the cursor, with recognition running entirely on the machine.
bolipi
Fireflies launches Talk, free unlimited voice dictation in its desktop app
A meeting-notes vendor undercuts $12–15/month standalone dictation by bundling unlimited desktop STT on Free — same company that shipped Voice Agents on 2 September (2026-09-02-fireflies-voice-agents).
Also Use
Details4 sources · 2 X posts
Fireflies shipped Talk inside its Mac and Windows desktop app: hold a hotkey, speak, and cleaned text lands at the cursor in any app, free and unlimited on every plan including Free, in more than 90 languages.
Fireflies Talk
FermionModel providermodel release2026-09-29
Fermion ships Phonon-2 open-weight English ASR at 164 MB
Shrinks open English ASR to a 164 MB CC-BY-4.0 download that Fermion says beats Whisper large on average WER and runs ~174× realtime on an M5 MacBook Air — vendor Open ASR Leaderboard scoring, one month after Phonon-1 (2026-08-28-fermion-phonon-1).
Details3 sources · 1 X post
Fermion released Phonon-2, a 164 MB English speech-recognition model that claims higher average accuracy than OpenAI Whisper large despite being about a tenth the download size, and transcribes about an hour of audio in roughly 20 seconds on a MacBook Air under CC-BY-4.0.
Phonon-2
Introducing Phonon-2: a new standard in speech recognition per byte. At just a tiny 164MB download, it is more accurate on average than OpenAI's Whisper large (a model 10X its size). Transcribe an hour of audio in just 20 seconds on a MacBook Air. Open weights, CC-BY-4.0.
https://x.com/yoitsmanan/status/2104990913886031993
AssemblyAI ships Universal-3.6 Pro Realtime for noisy agent calls
Short-response errors fall 45% on AssemblyAI's own 12,460-scenario agent bench (1.45% versus 2.65%) at held 3.5 Pro latency and $0.45/hr — the gain that matters on yes/no turns over noisy phone lines, with the benches still vendor-run rather than an independent audit.
Details1 source
AssemblyAI released Universal-3.6 Pro Realtime streaming STT for voice-agent and telephony audio, covering 32 languages at the same latency and $0.45/hr rate as Universal-3.5 Pro under model id universal-3-6-pro.
Universal-3.6 Pro Realtime
Modulate raises $25M led by Future Ventures to scale audio-native Velma
A dated company primary for the 2026 Modulate raise that an earlier sweep skipped as funding-only — Velma sits beside the talking stack as an understanding and supervision layer.
Details2 sources · 1 X post
Modulate announced $25 million in new funding led by Future Ventures, with Hyperplane and Lakestar participating, bringing lifetime funding stated in the release to $60 million for its Velma audio-understanding platform.
Velma
Speaktext-to-speech18
Smallest AI partners with Tenstorrent to run Lightning V2 TTS on-prem
Gives Lightning V2 a dated on-prem Blackhole path with usage-based pricing — useful when GPU-class spend or data residency blocks cloud TTS for bank, telco, or healthcare agents.
Details2 sources · 2 X posts
Smallest AI and Tenstorrent opened a production on-prem path that runs Lightning V2 realtime TTS natively on Tenstorrent Galaxy Blackhole servers, with usage-based pricing and no upfront hardware commitment.
Lightning V2 on Tenstorrent · vendor Tenstorrent
LokutorModel providermodel release2026-10-02
Lokutor releases Ito, streaming neural TTS for the ESP32-S3 microcontroller
With Oído (2026-09-30-lokutor-oido) a $5 chip gets both halves of a voice loop, but no board has run Ito yet and its own central estimate is slightly slower than real time; Cactus Whistle (2026-10-02-cactus-whistle) the same day targets phone-class CPUs, not microcontrollers.
Details5 sources · 1 X post
Ito is a 4.4M-parameter English TTS, 4.9 MB at int8, with two voices at 24 kHz that streams a 25 ms first chunk on an ESP32-S3 without cloud or NPU, verified bit-exact so far only in Espressif's QEMU emulator.
Ito
KittenMLModel providermodel release2026-10-02
KittenML releases KittenTTS 2, a ternary 1.7B speech LM with voice cloning
Cloning-grade TTS at 1.7B parameters runs on CPU thanks to ternary weights, but the licence is not Apache-2.0 like KittenTTS 0.8 (2026-02-19-kittenml-kitten-tts-0-8): companies above USD 1M revenue or funding need a separate grant, and the language count differs between the card (10) and the README (20).
Details4 sources
KittenML (Stellon Labs) published KittenTTS 2 weights and the renamed kittenml package, a 1.7B speech language model with 47 voices, zero-shot cloning, and inline emotion tags that runs on CPU, plus a hosted API.
KittenTTS 2
Index TeamModel providermodel release2026-10-02net-new
Index Team releases Index-Echo-S2ST-9B for voice-preserving dubbing
Open dubbing weights with explicit source-voice conditioning, which is speak plus a speech-translation front end, not a conversational agent that decides what to say.
Also Hear
Details2 sources · 1 X post
Index-Echo-S2ST-9B is an Apache-2.0 speech-to-speech dubbing package that takes Chinese or English speech and speaks a translation in English, Spanish, Japanese, or Chinese while conditioning on the source speaker.
Index-Echo-S2ST-9B
speak once, hear yourself in another language 🗣️🌍
Index-Echo-S2ST is a new open model that dubs your speech and keeps your voice: talk in English or Chinese, get it back in English, Chinese, Japanese or Spanish
▶️ on Spaces
https://x.com/HuggingApps/status/2106003866974065097
Hume sunsets TTS and EVI APIs on 13 November 2026
Builders on Octave TTS or EVI have a hard cutover with permanent data deletion after 13 November 2026 — plan migration and export before that window rather than treat the notice as soft guidance.
Also Think
Details2 sources
Hume's developer changelog states that its text-to-speech and Empathic Voice Interface APIs remain supported until 13 November 2026 at 12:01 a.m. EST, after which access ends and account data is permanently deleted.
TTS and EVI APIs
Broccoli ships Voice Studio for custom home-service agent voices
Home-service operators get CSR-clone and promptable voices inside the booking agent — brand continuity on after-hours lines without a separate TTS vendor project.
Also Use
Details2 sources
Broccoli opened Voice Studio so home-service phone agents can clone a CSR from a short recording or script, generate a voice from a one-line description, or pick from more than 30 platform voices, then A/B against the live voice before switching.
Voice Studio
SunoPlatformlaunch2026-10-01net-new
Suno opens Speech (beta), spoken audio with matching background music
Brings a consumer music product into the speak layer with one model that writes voice and score together, rather than a developer TTS API. Other 2026 Suno music drops stay out of this filing.
Details2 sources · 1 X post
Speech (beta) generates spoken audio and original background music as one track inside Suno on web and mobile, from an idea, a poem, or text you already wrote plus a voice and musical style.
Speech (beta)
OpenAIModel providereol2026-10-01
OpenAI deprecates tts-1, tts-1-hd, and gpt-4o-mini-tts, with removal on 6 January 2027
OpenAI names no standalone TTS successor, so cascaded pipelines that call it only for speech output must open a realtime session or switch vendors within about three months. Lands the same week as Hume's TTS and EVI sunset (2026-10-02-hume-tts-evi-api-eol).
Details1 source
OpenAI's text-to-speech models leave the API on 6 January 2027, and the recommended replacement is gpt-realtime-2.1-mini through the Realtime API rather than a new standalone TTS model.
tts-1, tts-1-hd, and gpt-4o-mini-tts
MicrosoftModel providermodel release2026-10-01
Microsoft ships MAI-Voice-2.1-Flash TTS at $15/1M characters, 150 ms e2e
Pairs with same-day Transcribe-2-Streaming (2026-10-01-microsoft-mai-transcribe-2-streaming) so Microsoft’s hear and speak SKUs share one Foundry/Vercel path at agent-grade latency.
Details4 sources · 1 X post
Microsoft AI released MAI-Voice-2.1-Flash, a low-latency sibling to MAI-Voice-2.1 for high-volume voice agents, with the same 23-language coverage and ~150 ms end-to-end latency for up to 45 seconds of audio.
MAI-Voice-2.1-Flash
Introducing 3 new models: MAI-Transcribe-2-Streaming, MAI-Voice-2.1 and MAI-Voice-2.1-Flash.
Accurate streaming transcription. Natural speech and less waiting between turns.
Build voice agents that keep the conversation moving!
https://x.com/MicrosoftAI/status/2105693024013467905
MicrosoftModel providermodel release2026-10-01
Microsoft ships MAI-Voice-2.1 multilingual TTS at $22/1M characters
Cross-language speaker consistency at 23 languages / 26 locales lets one brand voice stay native across English, Mandarin, German, and peers without swapping voices mid-session.
Details4 sources · 1 X post
Microsoft AI released MAI-Voice-2.1, its strongest multilingual text-to-speech model yet, covering 23 languages and 26 locales with a single speaker identity that keeps a native accent when switching languages.
MAI-Voice-2.1
Introducing 3 new models: MAI-Transcribe-2-Streaming, MAI-Voice-2.1 and MAI-Voice-2.1-Flash.
Accurate streaming transcription. Natural speech and less waiting between turns.
Build voice agents that keep the conversation moving!
https://x.com/MicrosoftAI/status/2105693024013467905
LiveKit Agents 1.8.4 adds Microsoft AI TTS, Eleven v4, and SmallestAI continuations
Wires same-day MAI-Voice-2.1/Flash and Eleven v4 into Agents while SmallestAI gains TTS continuations — builder path without waiting for a minor.
Details1 source
livekit-agents 1.8.4 adds a Microsoft AI speech plugin, ElevenLabs eleven_v4 and eleven_v4_turbo, SmallestAI TTS continuations protocol support, and a Deepgram TTS speed parameter.
livekit-agents 1.8.4
GradiumModel providermodel release2026-10-01
Gradium makes its sub-50 ms TTFA TTS the default after beta
Moves the under-50 ms TTFA path from `gradium-tts-beta` opt-in to the production default — Speko naturalness claim is vendor-cited, not re-run here.
Details2 sources · 1 X post
Gradium promoted its low-latency Text-to-Speech model out of beta as the default API and Studio path, stating around 50 ms time to first audio with hard-case pronunciation retained.
Gradium TTS
Our default Text-to-Speech model now has around 50ms time to first audio.
The model scores highest on naturalness for any sub-100ms model on Speko public benchmarks.
How much of your turn budget does TTS take?
https://x.com/GradiumAI/status/2105689201278664804
Deepgram self-hosted 261001 watermarks Flux TTS and drops Whisper
Self-hosted Flux TTS operators hit a hard watermarker requirement on 1 October while Whisper is removed as a supported path, so on-prem speak stacks need a coordinated Engine and Helm change rather than a quiet container bump.
Also Hear
Details1 source
Deepgram’s October 2026 self-hosted release 261001 requires Flux TTS watermarking, removes Whisper model support, aligns FIPS Engine metrics with standard images, and improves Japanese and German formatting.
Self-Hosted release 261001
Epilude Narrator reads any Mac text aloud with an on-device EP-1 voice or ElevenLabs v4 cloud voices
Eleven v4 (2026-09-28-elevenlabs-v4) reached a consumer reader two days after its release, with a 95 MB local voice as the unmetered fallback; in the same week Fireflies Talk (2026-09-29-fireflies-talk) added dictation to a meeting-notes app, while Epilude grew from dictation into notes and reading.
Details3 sources · 2 X posts
Epilude added Narrator to its Mac dictation and meeting-notes app: select text in any app or import a PDF, web page, or image, and a floating player reads it aloud and marks each word as it goes, either locally or through cloud voices.
Epilude Narrator · vendor ElevenLabs
Narrator is here.
Listen to any text on your Mac, using @ElevenLabs v4 voices or our own local text-to-speech model EP-1 voice.
Check out Narrator: https://t.co/Oc4pGeUMs4 https://t.co/hcrxl5RXVE
https://x.com/EpiludeHQ/status/2105392755480412413
Today we've introduced Narrator to @EpiludeHQ. Your Mac can now read any text to you, in voices you'll actually like to listen to, completely privately on your Mac.
We've tried to make it the smoothest listening and reading experience possible, and think we've found a form factor that works pretty well.
Our goal is to give every Mac owner the voice tools they need to be productive and get things done, which now include accurate dictation, meeting notes with Notetaker, and text-to-speech with Narrator.
All using Epilude's EP-1 family of models, and all processing completely privately on your Mac so you stay in control of your data.
Check it out:
https://t.co/shsAdXEc1E
https://x.com/axonbit/status/2105381794056937485
Deepgram adds Flux TTS inline pause and IPA pronunciation controls
Drug names and intentional silence land as first-class Flux TTS markup instead of Aura-style ellipsis prompting — pronunciation is still Early Access and cannot stack with pause or non-default speed.
Details3 sources
Deepgram shipped escaped pause markers and Early Access inline IPA pronunciation on Flux TTS (/v2/speak), alongside the existing speed control, with combination guardrails that reject degrading mixes.
Flux TTS inline pause and pronunciation
Telnyx Ultra adds seven Urdu and eight Odia TTS voices
South Asian Ultra coverage lands three days before the Chennai SIP anchorsite (2026-09-30-telnyx-chennai-sip-anchorsite), so India/Pakistan language TTS and India media PoPs arrive in the same W40 CPaaS push.
Also Connect
Details1 source
Telnyx Ultra now synthesises Urdu and Odia natively with fifteen new catalog voices usable from Voice AI Assistants, the TTS API, Call Control speak, and TeXML without a language hint.
Telnyx Ultra Urdu and Odia TTS voices
KyutaiModel providerother2026-09-28
Kyutai publishes Pocket TTS training via drifting one-step objective
Documents a Jacobian-free one-step recipe for the same Pocket TTS continuous-latent stack filed in W03 (2026-01-13-kyutai-pocket-tts) — vendor <1% WER / cloning claims, and an english_drifting_26-09 checkpoint path, without restating the January model launch.
Details4 sources · 1 X post
Kyutai published a technical blog on training the Pocket TTS sampler head with drifting, a one-step generative objective from Deng et al., claiming under 1% WER with voice cloning and calling it the first speech and first autoregressive model trained that way.
Pocket TTS drifting training
ElevenLabs ships Eleven v4 and low-latency v4 Turbo for expressive TTS
Puts a #1-ranked expressive TTS path and a ~100 ms Turbo SKU on the same architecture for ElevenAgents builders — vendor preference and latency figures are ElevenLabs' September 2026 benches, not an independent audit.
Details4 sources · 1 X post
ElevenLabs released Eleven v4 and Eleven v4 Turbo text-to-speech models with 90+ languages, richer inline delivery tags, and Instant Voice Clones from about 10 seconds of audio, with Turbo aimed at realtime voice agents.
Eleven v4 and Eleven v4 Turbo
Orchestrateframeworks, evals16
TelnyxPlatformlaunch2026-10-01
Telnyx opens Agent Memory API beta for per-caller recall across calls and chats
Caller memory moves into the CPaaS account that already owns the phone number, so a repeat caller can be greeted with facts from the last call; asynchronous beta writes mean a fact from a call that has just ended may take seconds to become recallable.
Details1 source
Telnyx put Agent Memory into beta: one API that stores facts extracted from voice calls, chats, or app sessions against a profile per user, caller, or agent, and returns ranked recall when the next conversation starts.
Agent Memory API (beta)
TCNPlatformacquisition2026-10-01
TCN acquires Abstrakt for real-time agent assist and automated QA
Puts live agent-assist and automated QA under the same cloud CCaaS roof as TCN's Virtual Agents — a contact-centre acquisition on the assist/QA layer rather than a speech-model buy.
Also Connect
Details2 sources
TCN acquired Abstrakt to fold real-time agent assist and automated quality assurance into its cloud contact-centre platform and workforce optimisation services.
Abstrakt acquisition
Phonely's October update adds outbound campaign controls and a transcript-to-agent beta
Phonely's dated release expands the platform from inbound agent building into outbound operations and a recording-to-agent workflow; the recording builder is explicitly beta and the changelog gives no independent quality or production-success evidence.
Also ConnectUse
Details1 source
Phonely's October platform update adds outbound campaign setup and queue controls, workflow and Ask AI improvements, and a beta that builds an agent from call recordings.
October 2026 platform update
OnepinPlatformlaunch2026-10-01net-new
Onepin launches a TTS production layer for pronunciation and line QC
Production TTS failures are often brand-name and number misreads, not model naturalness — Onepin sells a shared pronunciation and line-QC layer across 30+ engines so teams stop regenerating whole takes for one wrong word.
Also Speak
Details2 sources · 2 X posts
Onepin launched a production step after text-to-speech that checks every voiceover line against a 4-million-word pronunciation dictionary, normalises prices and dates, scores naturalness, clarity, and word accuracy, and fixes a single wrong word in the same voice without re-rendering the take.
Onepin
DeepgramModel providerpartnership2026-10-01
AWS Startups guides private-cloud Deepgram STT/TTS on SageMaker with Pipecat
Documents the three-party path for regulated private-cloud voice agents — Deepgram models, SageMaker bidirectional hosting, Pipecat orchestration — alongside same-week PolyAI Dialog-RSN-1 on SageMaker (2026-10-06-polyai-dialog-rsn-1-aws-sagemaker, W41) as a different Marketplace model listing.
Also HearSpeak
Details3 sources · 1 X post
An AWS Startups Insights post dated 1 October 2026 walks through deploying Deepgram speech-to-text and text-to-speech models on Amazon SageMaker AI for private-cloud voice agents, with Pipecat orchestration and SageMaker bidirectional streaming.
Deepgram on Amazon SageMaker AI with Pipecat · vendor AWS
If you're building voice agents and deploying on AWS, it's worth keeping up with the really good technical content that AWS solutions architects are publishing.
Here's a new blog post that walks through deploying Deepgram STT and TTS models in your private cloud on AWS.
Sagemaker is AWS's platform for GPU-based workloads. Deepgram models fully support Sagemaker deployments, and use a feature set called "Sagemaker bidirectional" for low-latency, streaming connections to the models.
We helped build the original Sagemaker bidirectional reference implementations in Pipecat, so there's excellent Pipecat support for these AWS capabilities.
https://x.com/kwindla/status/2107205079338270811
Decagon launches Duet Apprentice beta for learning from docs and Slack
Agent builders still paste SOPs by hand while policies drift in Slack — Apprentice closes that loop on the same Dialogues day as Voice 3, with a vendor claim that Duet already drafts more than 70% of live AOPs.
Details2 sources · 1 X post
Decagon introduced Duet Apprentice in beta so Duet can onboard from Notion, Confluence, SharePoint, or Google Drive, learn from escalated human-rep conversations, and keep updating Agent Operating Procedures from Slack and Microsoft Teams activity.
Duet Apprentice
Decagon launches Agent Modules for journeys beyond support
Concierge platforms that stop at ticket deflection leave lead-to-collections orchestration to custom builds — Modules package those journeys with Simulations, Campaign Composer, and shared memory on the same 1 October Dialogues stack as Voice 3.
Also Use
Details2 sources · 1 X post
Decagon introduced Agent Modules, journey-specific infrastructure, analytics, and Campaign Composer for lead qualification, onboarding, collections, and growth use cases that reuse one agent with shared memory across instances.
Agent Modules
Pipecat demo routes one pipeline across two to six Jev character workers
Multi-party voice agents usually mean multiple sessions; this pattern keeps one pipeline and lets Jev choose the speaker, which is the same classifier surface Pipecat 1.12.0 just shipped for evals and UI workers (2026-09-25-pipecat-1-12-0-classifiers-evals).
Also Speak
Details4 sources · 3 X posts
Jon Taylor (@JonPTaylor) shows a single Pipecat bot pipeline where a Typesafe Jev classifier picks which character worker should answer, then scales the same pattern from two workers to six, each with its own PhoneLLM prompt and TTS voice.
Multi-character Jev handover demo · vendor Typesafe
Weeks on and I'm still finding awesome ways to use @typesafeai Jev in Pipecat. This demo is ONE bot pipeline, with a classifier front-running the LLM to determine who the user is talking to. Pipecat then orchestrates a handover between two 'character' workers.
Oh, and of course, Opus 5.5 one-shotted this idea. Code might be ugly: https://github.com/jptaylor/pipecat-one-pipeline-two-characters
https://x.com/JonPTaylor/status/2105106940669915190
@typesafeai ... and why not 6 friends? (again, this is all just one pipeline, nuts!)
https://x.com/JonPTaylor/status/2105395457279070616
Self-organizing voice conversations. @JonPTaylor is building mind-expanding stuff with Jev and Pipecat.
This is one Pipecat process with six workers. Each worker is a "character" in the conversation, and has its own PhoneLLM system prompt and its own TTS.
You can talk to all of them, and Jev decides "who" should respond. The can also all "hear" each other, because all the workers share the conversation context.
https://x.com/kwindla/status/2106956074184261991
Hugging Face opens an objective Open TTS Leaderboard for multilingual and cloning models
Gives open TTS and cloning models a scalable public board beside arena Elo scores — useful for agent builders watching TTFA and multilingual WER, not a substitute for human preference.
Also Speak
Details1 source
Hugging Face launched the Open TTS Leaderboard, ranking open-weight text-to-speech and voice-cloning models with objective intelligibility, speed, and speaker-similarity metrics plus a Listen tab for human comparison.
Open TTS Leaderboard
VapiPlatformlaunch2026-09-29
Vapi launches Vapi CX to build, test, and grade voice agents for customer-experience teams
Vapi CX extends Vapi Model Intelligence (2026-07-21-vapi-model-intelligence) and traffic splitting (2026-10-12-vapi-version-traffic-split) into a build-to-observe loop; Kavak's 20-point NPS and 30% conversion changes are customer figures quoted by Vapi, and the 29 September blog predates the 21 October X launch.
Details2 sources · 1 X post
The CX toolkit gives customer-experience, operations, product, and engineering teams controls for building, deploying, observing, and optimising voice agents.
Vapi CX
A year ago, success with voice AI was a working demo.
Today, it's agents that resolve the hard calls, not just deflect the easy ones.
Clearing that bar takes depth for those building the agent and speed for the CX team running it.
That's why we're launching Vapi CX.
https://x.com/Vapi_AI/status/2112938908027617505
Simple AI introduces Tango, an audio-native model for end-of-turn and interruption decisions in its voice agents
Smart Turn from Pipecat is the named baseline, a month after Sesame open-sourced TurnBench (2026-08-28-sesame-turnbench); Tango's lead rests on Simple AI's own runs of the LiveKit benchmark, with no public weights or API to reproduce them.
Also Hear
Details2 sources · 1 X post
Simple AI trained its own conversational awareness model so its sales and support agents decide from the audio itself when a caller has finished and whether a cut-in is a real interruption.
Tango
Voice AI's biggest challenge is timing. Today we’re excited to introduce Tango, Simple's conversational awareness model.
Tango is industry-leading on LiveKit's public end-of-turn benchmark, catching 96% of turn endings, responding in ~800ms, and barging in 5% of the time.
For more: https://t.co/EglgPwDGtM
https://x.com/simpleailab/status/2104979521418264787
Hume AIModel providerother2026-09-29
Hume publishes a five-dimension multi-speaker conversational TTS evaluation
Same-gender speaker separation remains the failure mode overall scores hide — Gemini mixed-gender pairs average 4.65 while male-male averages 2.87, and ElevenLabs v3 flips the pattern (male-male 4.33, female-female 3.02) — so dialogue TTS buyers need this rubric beside cloning and controllability boards (2026-09-10-hume-ai-voice-replication-leaderboard, 2026-09-17-hume-voice-controllability-leaderboard).
Also Speak
Details2 sources · 1 X post
Hume published a bespoke multi-speaker dialogue rubric with 48 two-speaker scripts and blind human ratings across five dimensions, applying it to Gemini TTS generations and peer multi-speaker systems.
Multi-speaker TTS evaluation
SierraPlatformlaunch2026-09-28
Sierra turns Ghostwriter into a proactive Slack and Teams teammate for agent ops
Agent improvement loops that used to wait for a weekly call review now start from an always-on teammate that reads live voice and chat traffic — useful for Sierra operators who already run production agents, with the caveat that the post is still a controlled rollout rather than a numbered GA date.
Also Use
Details1 source
Sierra released a Ghostwriter update that lives in Slack and Teams, reviews customer-agent conversations, and proposes experiments without waiting for a prompt.
Ghostwriter
OcularPlatformlaunch2026-09-28
Ocular launches a full-duplex audiovisual conversational dataset
Video conversational and interaction models need four streams on one clock — speaker A/B audio plus video — and public full-duplex A/V sets remain thin (ViCo 1.6 h; ViCo-X 0.4 h in the INFP survey), so a 1,000+ hour on-request corpus is the video counterpart to Ocular's earlier speech Hi-Fi work.
Also HearSee
Details2 sources · 1 X post
Ocular released a full-duplex audiovisual dataset of two-person video calls with four synchronised streams — per-speaker audio and video on one shared clock — built for video conversational models.
Full-Duplex Audiovisual Dataset
Genesys ends support for the prior Agentic Virtual Agent model
Operators on the prior AVA model have a dated cutover window — test before auto-migration on 1 October 2026 rather than treat the notice as optional.
Also Think
Details1 source
Genesys Cloud's 28 September 2026 release notes state that support for the existing Agentic Virtual Agent model ends on 1 October 2026, with a 30-day opt-in window before automatic migration to the latest model.
Agentic Virtual Agent (prior model)
CekuraPlatformlaunch2026-09-28
Cekura opens a speech-to-speech phone-agent benchmark across nine realtime models
Long-form Medicare reliability separates the field far more than short clinic bookings — GPT Realtime 2.1 holds 61% pass³ on the long form while several peers fall under 15%, so teams picking a duplex model for multi-field phone intake get a shared, public call log instead of vendor demos alone.
Also Think
Details2 sources · 1 X post
Cekura published an open speech-to-speech benchmark that runs the same Pipecat phone agent on live calls across nine realtime voice models over 82 scenarios and three runs each.
Speech-to-speech benchmark
Connecttelephony, channels7
Pipecat ships C++ client SDK 1.0.0 rewritten for RTVI protocol 2.1
Native Unreal, robotics, and embedded voice clients can speak the same RTVI 2.1 session contract as the JS client without staying on untagged 0.x; the same week’s Unreal multi-NPC demo (2026-10-06-pipecat-unreal-multi-npc-demo, W41) embeds this C++ path.
Also Use
Details3 sources
The native Linux, macOS, and Windows client matches @pipecat-ai/client-js on RTVI 2.1, with a new PipecatClient, typed callbacks, Daily transport, and CMake/pkg-config installs, and it is not compatible with the earlier untagged 0.x builds.
Pipecat C++ client SDK v1.0.0
LiveKit Agents 1.8.4 adds Azure Voice Live Realtime API support
Azure Voice Live becomes a first-class Agents realtime path the same day MAI ships Transcribe-2-Streaming and Voice-2.1 on Foundry.
Details1 source
livekit-agents 1.8.4 adds Azure Voice Live Realtime API support and forwards Twilio CallToken during warm transfers.
livekit-agents 1.8.4
Decagon launches Personal Agent Gateway for Muse-class personal agents
With Muse, dots, and Instinct already placing brand contacts, Decagon’s 1 October gateway is a concrete connect path for agent-to-agent CX with detection and scoped authorisation instead of hoping personal agents pass as people.
Also OrchestrateUse
Details2 sources
Decagon launched Personal Agent Gateway so enterprises can detect customers’ personal AI agents and serve them on a dedicated channel with explicit authorisation scopes rather than treating them as ordinary human callers.
Personal Agent Gateway
TelnyxPlatformlaunch2026-09-30
Telnyx opens a Chennai SIP anchorsite as India’s second local voice PoP
Gives Indian and India-serving voice AI stacks a second domestic SIP media anchor (Mumbai + Chennai) so southern-origin calls need not hairpin through a single national PoP before the Telnyx backbone.
Details2 sources · 1 X post
Telnyx added a Chennai, India anchorsite for SIP connections so voice traffic can enter its private backbone locally beside Mumbai, giving contact centres and developers a second Indian entry point for redundancy and southern-India latency.
Chennai SIP anchorsite
Bland becomes an official Twilio technology and go-to-market partner
Packages Bland's self-hosted voice-agent stack with Twilio Super Network rails that compliance teams already know — BYO account keeps existing numbers and rates, so the 30 September wire is distribution and trust more than a new telephony API.
Also Use
Details2 sources
Bland announced a technology and go-to-market partnership with Twilio so every Bland customer gets Twilio-powered calling and number provisioning built in, and existing Twilio customers can bring their own account.
Twilio technology and GTM partnership · vendor Twilio
wevnalPlatformpartnership2026-09-29net-new
wevnal and SUPER STUDIO link BOTCHAN AICALL with ecforce for e-commerce phone service
The companies aim for a 100% answer rate. A 2026 wevnal survey cited in the release found that 88.7% of respondents saw a quality gap between experienced and new operators, while 40.5% expected AI to reproduce experienced operators' know-how, compared with 21.4% expecting cost cuts. No commercial terms were given.
Also Use
Details3 sources
The agreement covers API integration, joint phone-service work, and support for ecforce merchants.
BOTCHAN AICALL × ecforce alliance
Delight.ai launches IVR Navigation for voice-agent transfers
The feature targets a transfer failure point between two organisations' phone systems, allowing an agent to carry context through a menu and human handoff instead of abandoning the caller at 'press 1'.
Also OrchestrateUse
Details2 sources · 1 X post
Delight.ai's IVR Navigation lets a voice agent work through a mapped phone menu, wait for a person, and brief that person before the caller joins.
IVR Navigation
Introducing IVR Navigation.
Nobody likes sitting through "For billing, press 2." Now your AI does it for you.
A Delight voice agent works through the phone menus you've mapped, waits for a person to pick up, and briefs them before the caller joins. The caller never hears a single beep.
Full breakdown in the comments.
https://x.com/GetDelightAI/status/2104770871026233441
Useagents in market31
SponklyPlatformlaunch2026-10-04net-new
Sponkly translates chats, calls, and video meetings live, with subtitles and a synthetic voice in each listener's language
ElevenLabs aims Eleven Realtime Translation (2026-10-21-elevenlabs-eleven-realtime-translation, W43) at telecom operators and contact centres, while Sponkly sells the same person-to-person job to consumers and small teams alongside Coretech Labs' Nani? subtitles (2026-10-16-coretech-labs-nani-live-meeting-subtitles, W42) (analytical). The maker's 2–4-second figure on Product Hunt is slower than the site's 1–2 seconds, and no model vendor is disclosed.
Also SpeakHear
Details5 sources · 2 X posts
People who do not share a language can message, call, or meet on the web or Android while each reads and hears the others in their own language, with end-to-end encrypted chats.
Sponkly
Credgenics launches Prix AI for outcome-priced collections voice agents
Indian collections SaaS turns a billion-call corpus into an outcome-priced voice platform — sub-800 ms and Bucket X paid-rate claims are vendor-stated, and the company newsroom index still lacked the wire at capture.
Also Orchestrate
Details4 sources
Credgenics opened Prix AI, an enterprise voice-agent platform for collections and broader BFSI conversations, trained on insights from more than 1 billion collections calls across over 100 million loan accounts.
Prix AI
Liberty Home Guard puts Aria and Ari voice agents on its claims and technician lines
Voice automation reaches the technician side of a home-warranty claim, with labour and parts authorisation plus a live human handoff on the same call; the 2 October release names no vendor and publishes no containment or cost figures.
Details2 sources
Liberty Home Guard launched two AI voice agents, Aria for homeowner claims and Ari for service-technician diagnostic authorisation, live on its existing phone lines.
Aria and Ari AI voice agents
DecagonPlatformpartnership2026-10-02
Decagon joins OpenAI Marketplace as a launch partner for CX agents
OpenAI commitment spend as a buy path for Decagon agents — distribution, not a new voice SKU, three days after Voice 3/Chord at Dialogues.
Also Orchestrate
Details2 sources · 2 X posts
Decagon became a launch partner on OpenAI's B2B marketplace so eligible enterprise customers can apply existing OpenAI commitments toward purchasing Decagon customer-facing agents.
OpenAI Marketplace · vendor OpenAI
Banner Health puts ElevenAgents on primary-care scheduling calls around the clock
A top-tier US nonprofit health system is letting a voice agent write appointments into the EMR on the first call, with warm transfer and HIPAA governance as the safety rail; the blog gives no volumes or go-live metrics, and the 9 October ElevenLabs post is the public X amplification of the 2 October story.
Also Connect
Details2 sources · 1 X post
Banner Health's ElevenAgents answer inbound primary-care calls with no hold queue, book, reschedule, or cancel appointments in the EMR, and warm-transfer to a Banner team member with call context when a person is needed.
ElevenAgents primary-care scheduling · vendor ElevenLabs
We're partnering with Banner Health to answer patient calls with AI voice agents, starting with primary care scheduling. Patients call any time of day or night, and ElevenAgents books, reschedules, or cancels appointments directly in Banner's electronic medical record. When a patient needs a person, the call transfers to a Banner team member with the necessary context. Read more: https://elevenlabs.io/blog/banner-health
https://x.com/ElevenLabs/status/2108580093224518133
Allstar Services extends Revin voice and SMS agents to ten more roofing brands
A PE-backed roll-up is standardising first contact on one voice-agent vendor across acquired local brands while keeping ten distinct agent voices; the 30% unassisted-booking figure is Allstar's own and comes from one brand over about eight months.
Details3 sources
Residential roofing platform Allstar Services is rolling Revin's AI voice and SMS agents out to ten more of its 19 regional brands across six Midwest and East Coast states, each configured to its own brand rules and voice.
Revin AI voice and SMS agents · vendor Revin
ResiDesk adds Voice AI for Leasing to answer multifamily prospect calls
Phone becomes the fourth channel in a multifamily AI stack that already ran text, email, and web chat, so a prospect call can book a tour with the same context later used for renewals; the release gives no call metrics, and its 10% renewal uplift is a platform-wide average rather than a voice result.
Also Connect
Details4 sources
ResiDesk, an AI operations platform for multifamily owners and operators, made Voice AI for Leasing available, answering inbound prospect calls around the clock with live availability, pricing, and tour booking into the property calendar.
Voice AI for Leasing
RealVoice Travel Group launches TripBeast Voice Agent for vacation-rental phone bookings
Card capture moves to an SMS link rather than the voice line, a PCI-friendly pattern for rental managers; no containment or booking-conversion numbers accompanied the 4–6 October VRMA debut, so outcomes remain unproven.
Details3 sources
RealVoice Travel Group commercially launched TripBeast Voice Agent, a phone agent that lets vacation-rental companies search inventory, book or cancel stays, and text secure payment links to callers at any hour.
TripBeast Voice Agent
PayNearMe launches an AI Servicing and Collections Agent in PayXM
Fintech-embedded outbound voice for collections/servicing — pilot throughput (~28 staff-hours of dials in one hour) is vendor-claimed from Indiana Finance Company, not an independent audit.
Also Connect
Details3 sources · 1 X post
PayNearMe shipped a voice-and-text agent inside PayXM that places outbound calls, answers inbound, and helps customers make or schedule payments, escalating to staff when needed.
AI Servicing and Collections Agent
Today, we're launching the AI Servicing and Collections Agent, a new agent built directly into the PayXM™ platform to expand inbound self-service and outbound engagement.
Read the announcement: https://home.paynearme.com/press/paynearme-launches-ai-servicing-and-collections-agent-to-help-support-teams-scale-while-lowering-the-cost-to-serve/
https://x.com/PayNearMe/status/2105634124904227324
EvaSpeaks launches an omnichannel AI agent platform for leads and appointments
SMB-facing omnichannel voice+email agent with self-serve signup sits beside same-week inbound/outbound peers Rabona (2026-09-28-rabona-blended-ai-call-centre) and PayNearMe (2026-10-01-paynearme-ai-servicing-collections-agent) — similarity of agent use layer only, not a partnership.
Also Connect
Details2 sources
EvaSpeaks opened a self-serve omnichannel AI agent platform that turns phone, web, email, and outbound conversations into qualified leads, booked appointments, CRM updates, and follow-up actions.
Omnichannel AI Agent Platform
ElevenLabs gains FedRAMP 20x Class A for ElevenAgents and speech APIs
US federal procurement for ElevenAgents and speech APIs can start from a published FedRAMP Class A package rather than a greenfield review, a day after the $22B tender priced the agents business.
Also SpeakHearOrchestrate
Details1 source · 1 X post
ElevenLabs published FedRAMP 20x Class A certification covering ElevenAgents plus Text to Speech and Speech to Text APIs in Zero Retention Mode with US data residency, extending ElevenLabs for Government to US federal buyers.
ElevenLabs for Government
BT Business launches AI Receptionist for UK small firms under AI from BT
A UK incumbent carrier now resells a RingCentral-branded receptionist agent (vendor line of 2026-03-10-ringcentral-air-pro) to SMBs on numbers they already hold; the £3.7 billion missed-call figure is BT-commissioned Censuswide survey modelling, not measured loss.
Also Connect
Details5 sources
BT Business launched AI from BT, a UK business AI portfolio led by AI Receptionist, which answers calls 24/7, books appointments, and transcribes conversations on a firm's existing fixed or mobile number.
AI from BT · AI Receptionist · vendor RingCentral
AudiblePlatformlaunch2026-10-01net-new
Audible lets listeners talk to Renfield in a real-time Interactive Story beta
A consumer audio platform with a large audiobook catalogue is putting a speaking AI character inside a licensed production; the beta is limited to US and UK listeners and one title, with the second (Exoplanet, 5 November) already scheduled.
Also ThinkSpeak
Details3 sources
Audible opened a limited US and UK beta of Interactive Story, a spoken real-time conversation with a character from the Audible Original Bram Stoker's Dracula, beside a Character Guide and Visual Explorer.
Interactive Story
VapiPlatformlaunch2026-09-30
Vapi launches Personal Agent Calling so consumer assistants can place phone calls
Consumer-agent telephony moves from Muse/Instinct-class product features into a shared voice layer underneath many assistants, with a dated beta and a worked 14-minute hold example on 30 September.
Also Connect
Details1 source
Vapi opened Personal Agent Calling in beta at phone.vapi.ai, letting personal assistants such as Muse, Instinct, ChatGPT Work and Claude Code dial businesses, hold the conversation and report results on the user’s Vapi minutes.
Personal Agent Calling
RiachueloEnterprisepartnership2026-09-30net-new
Riachuelo runs Genesys Agentic Virtual Agent on WhatsApp for Brazilian retail CX
Puts a major Brazilian retailer on Genesys AVA with an 84% vs 30% self-service jump on WhatsApp — still a digital channel go-live, with voice journeys only described as future exploration.
Also Orchestrate
Details3 sources · 1 X post
Genesys said Brazilian fashion retailer Riachuelo is using Genesys Cloud Agentic Virtual Agent on WhatsApp Business so shoppers can resolve delivery and support issues without a human, after earlier Genesys Cloud Copilot for employees.
Genesys Cloud Agentic Virtual Agent (WhatsApp) · vendor Genesys
Marriott expands Cresta AI Agent for guest voice and chat answers around the clock
Puts a major hotel group’s guest-facing voice+chat path onto Cresta AI Agent as an expansion of an existing relationship — still Cresta’s X primary without a Marriott wire, so scope and metrics stay thin.
Also Connect
Details2 sources · 1 X post
Cresta said Marriott International is expanding their relationship so Cresta AI Agent handles guest inquiries across voice and chat inside Marriott’s patent-pending tech platform, giving faster multilingual answers while associates keep higher-touch work.
Cresta AI Agent (guest experience expansion) · vendor Cresta
MallowPlatformlaunch2026-09-30net-new
Mallow puts its screen-free AI voice game speaker for children on sale in France
The child's voice is the only input on a €119.99 device, which puts child-speech ASR and guardrails in the critical path; Mallow keeps its model stack undisclosed and told Frandroid half of it may change by February.
Also Hear
Details4 sources
Mallow, the Paris start-up co-founded by Flore Cousin and Cédric O, launched a screen-free speaker on which children aged 4 to 10 play generative-AI audio adventures, quizzes, and mini-games by voice after placing an NFC figurine on top.
Mallow screen-free audio game speaker
Kia launches the Sorento in India with a SoundHound generative voice assistant
In-car generative voice ships on a production SUV in India a week after SoundHound's OASYS Edge architecture launch (2026-09-24-soundhound-oasys-edge), but the release does not say whether the Sorento runs on-device or in the cloud, or which Indian languages it supports.
Also Think
Details3 sources · 1 X post
The Kia Sorento launched in India carries the Kia AI Assistant with Real-Time Generative AI on SoundHound's voice platform, answering 'Hey Kia' with multi-turn control of cabin systems, owner's-manual questions, and live search and navigation.
Kia AI Assistant with Real-Time Generative AI (Sorento) · vendor SoundHound AI
jambuild launches voice-driven multiplayer vibecoding on Product Hunt
A concrete end-user voice coding product that names Deepgram + Pipecat in the stack without being a Pipecat employee demo — filed as launch/use after an earlier W40 open-check skip that treated it as voice-input-only.
Also OrchestrateHear
Details4 sources · 2 X posts
Product Hunt CEO Rajiv Ayyangar launched jambuild, a browser multiplayer coding harness where two people talk and point to ship web-app changes in seconds, live on Product Hunt for Hypership day.
jambuild
I'm launching jambuild: fast multiplayer vibecoding powered by voice
Like @aaron_epstein and others, I've been searching for a truly magical multiplayer agent experience. I recently realized, in thinking about it, that multiplayer agentic coding is no fun if you have to wait for the agent, so I made an ultra-fast voice-driven harness.
I'm trying to recreate the flow of jamming with someone on @figma or on a whiteboard, but have the medium be code.
It uses @DeepgramAI and @pipecat_ai and a small model I trained to do pre-emptive turn detection, and non-thinking coding models with some speed optimizations.
It's live on Product Hunt, and since today is Hypership day, I'll get extra points if I build features you ask for. So...try it out and ask me for features!
https://x.com/rajivayyangar/status/2105307326416908677
https://x.com/rajivayyangar/status/2105307326416908677
https://x.com/kwindla/status/2105376519858299120
ElevenLabs closes a $300M employee tender at a $22B valuation
A liquidity tender at $22B, not a primary raise, still prices ElevenAgents-led growth at twice the February Series D after the company already shipped Eleven v4 the same week.
Also SpeakOrchestrate
Details2 sources · 3 X posts
ElevenLabs completed a $300 million employee tender offer led by Wellington and T. Rowe Price that values the company at $22 billion, double its February Series D mark, citing enterprise demand for ElevenAgents.
ElevenAgents
call4mePlatformlaunch2026-09-30net-new
call4me launches an MCP server that lets Claude Code, Codex, and other agents phone businesses
call4me shipped the same day as Vapi Personal Agent Calling (2026-09-30-vapi-personal-agent-calling) and five days before ClawCall's ChatGPT plugin (2026-10-05-clawcall-chatgpt-plugin, W41); by 8 October the founder said Muse drove about ten times the use of Instinct or Grok Bot, so a third-party calling tool found Muse users even after Muse's own outbound-calls beta (2026-09-17-meta-muse-outbound-calls, W38). All usage figures are founder-reported.
Also Connect
Details4 sources · 2 X posts
Users of terminal coding agents and assistants such as Muse and Grok Bot can hand a phone errand to call4me, which dials the business and returns the result to the agent.
call4me
introducing call4.me, an ai agent that handles phone calls for you.
you can use it to cancel subscriptions, book medical appointments, dinners, change flights, or anything else possible with a call.
most importantly, it works in claude code & codex. i spend most of my time in the terminal, so i want my productivity tools to live there as well. switching to imessage or another app annoys me.
i've now handled dozens of tasks over the phone using it & am in love. if you're like me & hate breaking concentration to dial, you'll likely love it as well.
https://call4.me/
https://x.com/skeptrune/status/2105320262938009690
call4me is live on producthunt: https://www.producthunt.com/products/call4me !!
it's the best way to let your agent handle real world tasks.
to date, there have been 962 calls placed with hundreds of successful outcomes. i've filled the examples page & blog up with some of the coolest ones: https://call4.me/examples
the most popular integration has been muse by far. a few people have tried it from instinct & grok bot, but about an order of magnitude less than muse.
building a mcp server that adds some non-native functionality to an agent seems like free money in 2026.
https://x.com/skeptrune/status/2108230748629356845
SierraPlatformpartnership2026-09-29
Sierra joins OpenAI Marketplace as a launch partner for enterprise agents
Distribution via OpenAI spend commitments, not a new voice SKU — sits beside same-week Ghostwriter teammate (2026-09-28-sierra-ghostwriter-teammate) and W39 Liberty Global×Sierra (2026-09-23-liberty-global-sierra-partnership).
Also Orchestrate
Details2 sources · 1 X post
Sierra became a launch partner in OpenAI’s enterprise Marketplace so eligible customers can build Sierra agents and apply existing OpenAI commitments toward the purchase.
OpenAI Marketplace · vendor OpenAI
Proud to be a launch partner in @OpenAI's new marketplace! Now, eligible OpenAI customers can start building with Sierra, using existing OpenAI commitments.
We're excited to deepen our partnership with OpenAI, and make it even easier for businesses to build better customer
https://x.com/SierraPlatform/status/2104993581400514967
EliseAI raises $350M at a $4B valuation
This is a dated financing filing for the existing property-management and healthcare voice platform, separate from Apollo and the later Nextech preferred-partner announcement.
Details1 source · 1 X post
EliseAI announced $350 million in financing at a $4 billion valuation, led by Andreessen Horowitz, Bessemer Venture Partners, and Ontario Teachers’ Pension Plan.
ElevenLabs lists ElevenAgents on the OpenAI Marketplace as a launch partner
Same DevDay distribution channel as Sierra (2026-09-29-sierra-openai-marketplace) for a TTS/agent vendor that shipped Eleven v4 the day before (2026-09-28-elevenlabs-v4).
Also SpeakOrchestrate
Details2 sources · 1 X post
ElevenLabs joined OpenAI’s enterprise Marketplace as a launch partner so eligible customers can access ElevenAgents and apply OpenAI commitments toward purchases once buying opens.
ElevenAgents OpenAI Marketplace · vendor OpenAI
ElevenAgents is now on the OpenAI Marketplace.
If your company is an OpenAI enterprise customer, you'll soon be able to give it a voice in 90+ languages and put it on your existing OpenAI commitment.
Learn more: https://elevenlabs.io/blog/elevenagents-now-on-the-openai-marketplace
https://x.com/ElevenLabs/status/2105004567335420258
CarMaxEnterpriselaunch2026-09-29
CarMax extends Sierra voice agents to every inbound store and customer experience centre call
Coverage went from inbound sales calls in the August release (2026-08-06-carmax-sierra-inbound, W32) to 100% of store and centre calls within about five months of a May go-live, and it surfaced on an earnings call rather than a vendor wire, which is the hunt-4 path enterprise rows keep arriving by.
Details3 sources
CEO Keith Barr told investors that AI voice agents now take all inbound calls to CarMax stores and its customer experience centres, answering what they can and routing the rest to the right associate.
Inbound store and customer experience centre voice agents (Sierra) · vendor Sierra
CactusPlatformlaunch2026-09-29net-new
Cactus adds marketing management to its home-services AI call and booking platform
Cactus moves from answering and booking into campaign attribution and estimate follow-up, the lead-reactivation work Netic sells to American Residential Services (2026-08-10-american-residential-services-netic-partnership); this home-services Cactus is a different company from Cactus Compute.
Also ConnectOrchestrate
Details2 sources
Cactus expanded its home-services AI platform with marketing management that connects campaign data to AI-handled calls, appointment outcomes, estimate follow-up, and customer reactivation.
AI call handling, appointment booking, and marketing management
Rabona AI opens a blended AI call centre that switches inbound and outbound on one agent
Japan-market inbound/outbound blend with named go-lives — Houmiya at ~90% AI-completed post-install inbound and ZAP at ~80% fewer delivery-date calls — plus an outcome-priced SKU for some accounts from 28 September.
Also Connect
Details3 sources
Tokyo University spinout Rabona AI began offering a blended AI call centre in which one agent handles inbound and outbound phone work with CRM context, post-call write-back, and SMS, citing Houmiya Equipment and ZAP as early operators.
Blended AI Call Centre
PrestoPlatformfunding2026-09-28
Presto raises $10M from Remus Capital affiliates to scale QSR Voice AI
A dated primary for a QSR drive-thru vertical already live with national brands, with Toast and ElevenLabs named in the same release window rather than a greenfield demo raise.
Details2 sources
Presto Phoenix announced $10 million from Remus Capital-affiliated investors and other existing backers to expand Presto Voice drive-thru deployments, product work, and QSR customer impact.
Presto Voice
NOSEnterprisepartnership2026-09-28net-new
NOS tests ElevenLabs-voiced support agents with 3-5% of its Portuguese customers
The 50% handled rate uses no-callback as a proxy for resolution and comes from a 3-5% test cohort, so it is an early figure; NOS buys only the voice from ElevenLabs, whereas Banner Health runs the full ElevenAgents platform on scheduling calls the same week (2026-10-02-banner-health-elevenagents-scheduling).
Also Speak
Details1 source
Portuguese telecom and media group NOS replaced static IVR menus with its own voice agents for informational calls, using ElevenLabs Text to Speech for the European Portuguese voice.
European Portuguese customer-service voice agents · vendor ElevenLabs
Liberate launches AI Intercept to route consumer AI-agent calls away from insurer staff
Consumer agents that place calls, such as Meta Muse outbound (2026-09-17-meta-muse-outbound-calls), now have a named receiving-side product in insurance, though Liberate gives no detection accuracy and no carrier reference for Intercept.
Also Connect
Details4 sources
Liberate's AI Intercept screens inbound calls to property and casualty carriers and agencies, detects when the caller is a consumer's AI assistant, and hands that call to Liberate's own agent for quoting, servicing, and follow-up.
AI Intercept
Instinct raises $1B Series C at a $10B valuation for its personal AI agent
The financing is distinct from Instinct Concierge's 16 September calling launch and from the 7 October Sierra/Decagon customer-service partnership; it supplies a separate corporate-development marker for the personal-agent company.
Also Orchestrate
Details2 sources
Instinct raised $1 billion in Series C funding at a $10 billion valuation to expand its personal AI agent.
$1B Series C