ML Data & Platform Engineer

The future of AI is nameless: Tresic selects Speechmatics to power the speech layer of its conversation intelligence platform. End customers will only ever see their own provider's brand, which means the layer underneath has to be good enough to go unnoticed. LONE TREE, Colo. and CAMBRIDGE, United Kingdom, Sept. 09, 2026 (GLOBE NEWSWIRE) - Tresic has selected Speechmatics to power the speech-to-text layer of its Intelligence Cloud platform, following a direct evaluation against Google and Deepgram. The platform, built on the vCon standard, adds conversation intelligence to calls a business is already making. Speechmatics runs as part of Tresic's own infrastructure, so audio and transcripts never leave it. "Every summary and search result depends on knowing who spoke and what they actually said, so diarization and accuracy carry more weight here than they would..." "Our partners have spent years earning trust with these customers, and it is their name that goes on this. That decided how we chose the speech layer..." "Every summary and search result depends on knowing who spoke and what they actually said, so diarization and accuracy carry more weight here than they would..." "Our partners have spent years earning trust with these customers, and it is their name that goes on this. That decided how we chose the speech layer..." Both companies sell into products that carry somebody else's brand. Tresic goes to market through communications and managed service providers. Speechmatics supplies the speech layer inside voice products that other builders and enterprises ship. A business deals with its own provider throughout and never has to think about who built what. That is becoming the standard route for business AI. From Slackbot to QuickBooks to Zoom's AI Companion, and most recently Apple Intelligence, the same shape keeps appearing through 2026: a familiar brand on the surface, somebody else's models behind it. Gartner expects it to stay that way. With AI in what the firm calls the Trough of Disillusionment through 2026, it says AI will most often reach enterprises through their incumbent software provider rather than arrive as a new project. How it works Tresic attaches to the voice seats a provider already sells. Nothing is installed on the customer's side, nobody invites a bot into a call, and staff answer the phone the way they always have. Where the seat is already recording, every analyzable business conversation produces a record. The platform is never in the call path: it cannot drop a call, add delay, or affect call quality. From that record, the business gets a summary the same business day, and every commitment becomes a task with an owner and a due date. When a conversation meets conditions the business has set, a manager is alerted within minutes of hang-up. Conversation history is searchable by person, number, and topic. "Our partners have spent years earning trust with these customers, and it is their name that goes on this. That decided how we chose the speech layer. Speechmatics does one of the hardest parts of the job and takes none of the credit, which is the same arrangement we offer our partners." Kevin Nethercott, Co-Founder and CEO, Tresic Why Speechmatics was selected Summaries, alerts, and search all rest on getting the words right, and redaction shows the stakes most clearly. A card number spoken aloud can only be masked if the system heard it correctly, and one wrong digit means it is not masked at all. Tresic names PCI redaction as the foundation layer of its platform, which puts transcription accuracy in the category of compliance controls. The audio makes that harder than it sounds. These are calls from vans, depots, shop floors, and open-plan offices, in accents from everywhere, with nobody reading from a script. Consistency across the test set and accuracy on exactly that kind of audio decided the evaluation. Deployment was the second condition. Because partners badge the platform as their own, conversation data has to stay inside infrastructure they can account for to their customers, so both passes run on Tresic's own hardware: one tuned for speed, one tuned for accuracy, so the record is right. "Every summary and search result depends on knowing who spoke and what they actually said, so diarization and accuracy carry more weight here than they would on cleaner audio. Running inside Tresic's own infrastructure costs them nothing in accuracy, which is usually the trade people expect to make. The version that works is the one nobody has to think about." Katy Wigdahl, CEO, Speechmatics All of it rests on vCon, the open IETF-track standard for conversation data, which keeps a provider's conversations portable and readable outside the platform. Data is separated by partner, customer, and location. About Tresic Tresic adds conversation intelligence to the voice seats communications and managed service providers already sell, so every business conversation ends in a summary, a captured commitment, or an alert to the person who needs it. Partners carry it under their own brand, at their own price, to the customers they already serve. The platform is built on vCon, the open IETF-track standard for conversation data, and is distributed through the channel only. Tresic is headquartered in Lone Tree, Colorado. Intelligence, Actioned. Visit tresic.cloud. Media Contact: Jennifer Singer, Chief Commercial Officer, [email protected] About Speechmatics Speechmatics builds foundational speech technology for file transcription, real-time captioning, and voice agent applications. Its APIs help global businesses understand every voice across languages, accents, dialects, demographics, and locations. Processing over 500 years of audio globally each month in 55+ languages, Speechmatics is headquartered in Cambridge, UK.

Speechmatics has launched Linden, a speech-to-text model designed for voice agents, now available through LiveKit Inference across over 55 languages. Developers can integrate the model in under a minute via UI or code. The model addresses accuracy issues in real-world conditions where voice agents struggle, including strong accents, noisy environments, and alphanumeric sequences over poor-quality phone lines. Linden specialises in transcribing non-native speakers, dialects, and account numbers. LiveKit Inference manages the pipeline, routing, billing, and turn detection, eliminating the need for separate API keys or accounts. Norwegian AI company boost.ai already deploys both technologies in production for European enterprise clients. The partnership aims to reduce deployment time from weeks to minutes whilst improving transcription reliability in challenging production scenarios.

Speechmatics and LiveKit target the accuracy gap breaking voice agents in production. Linden, Speechmatics' new speech-to-text model, is available now through LiveKit Inference, enabling developers to build reliable voice agents at scale across 55+ languages. CAMBRIDGE, United Kingdom, Aug. 27, 2026 (GLOBE NEWSWIRE) - From today, developers building voice agents on LiveKit Inference can switch to Speechmatics in less than a minute through the UI or code. Linden, the first Speechmatics model purpose-built for voice agents, is available through LiveKit Inference, LiveKit's platform for using leading voice AI models from one place. That means no separate API key, account or invoice. "LiveKit Inference exists so developers can ship and iterate fast, without the headache of managing API keys or rewriting their stack every time a model..." "What separates an agent people trust from one they find exhausting is whether it can understand what you said. Our new model solves for this. Linden not only..." "Voice agents live or die on the STT layer. Speechmatics gives us what actually matters in production: low-latency real-time models, fast iteration on the..." "LiveKit Inference exists so developers can ship and iterate fast, without the headache of managing API keys or rewriting their stack every time a model..." In a demo scenario, a voice agent typically hears one speaker, clean audio and a short exchange where all STT providers perform well. Real-world production is rarely that tidy: a caller with a strong accent, a noisy cafe, a 16-digit card number read out over a low quality phone line. When the transcript breaks on the words that matter, everything downstream breaks with it. Together, LiveKit and Linden solve that problem for the builder. Linden gets the words right where it is hardest: 55+ languages, non-native speakers and strong accents, and alphanumerics such as account and card numbers read out over a bad line. LiveKit Inference handles the pipeline, the routing, the billing and the turn detection that decides when the agent speaks. Adopting a speech provider used to take weeks, now it's something you can simply select. "What separates an agent people trust from one they find exhausting is whether it can understand what you said. Our new model solves for this. Linden not only excels at 55+ languages, but also transcribing non-native speakers, dialects, and accents. LiveKit built the infrastructure that made those deployments possible, and its community has worked out what production voice AI actually takes, in the real world, faster than anyone else." Ricardo Herreros-Symons, CRO, Speechmatics Norwegian conversational AI company boost.ai already runs Speechmatics and LiveKit in production across the voice agents it builds for European enterprises, on infrastructure that also powers xAI Grok and Salesforce Agentforce. "Voice agents live or die on the STT layer. Speechmatics gives us what actually matters in production: low-latency real-time models, fast iteration on the hard edges like alphanumerics and end-of-utterance, and serious language breadth across Europe. Just as importantly, the people behind it are a pleasure to work with; close, hands-on, and genuinely invested in getting things right." Rasmus Hauch, CTO, boost.ai LiveKit Inference, launched in October 2025, lets developers build a voice agent in minutes. A single API key calls the most popular STT, LLM and TTS models, and models swap with a one-line code change. Consolidated billing, rate-limit management and usage reporting all run through LiveKit Cloud. Speechmatics is available natively through LiveKit Inference, via the Speechmatics plugin for LiveKit Agents Open Source framework or through LiveKit's Agent Builder. "LiveKit Inference exists so developers can ship and iterate fast, without the headache of managing API keys or rewriting their stack every time a model changes. Adding Speechmatics equips voice agents with exceptional speaker diarization that clearly tracks who said what with any number of speakers from day one. This capability is especially powerful for physical AI and real-world multi-speaker environments." David Zhao, CTO & Co-founder, LiveKit About Speechmatics: Speechmatics builds the most accurate and inclusive speech recognition in the world, trusted by enterprises across financial services, media, contact centers and government for unmatched language coverage, accuracy, and flexibility. About LiveKit: LiveKit is the open source framework and cloud platform for voice, video, and physical AI agents. Used by developers and enterprises worldwide, LiveKit powers realtime, multimodal AI applications at scale - from customer support and telephony to robotics and consumer products. The company is headquartered in San Francisco. Learn more at livekit.com. Media Contact Mieke Smith // [email protected]