Voice AI Agent Exit: Salesforce to Acquire Tenyx

Salesforce has signed a definitive agreement to acquire Tenyx, a developer of voice AI Agents that are redefining how enterprises engage with their customers through natural, human-like conversational AI experiences.

The acquisition of Tenyx is expected to enhance Salesforce’s Agentforce Service Agent by incorporating Tenyx’s three core AI technology pillars: a robust conversational voice platform, cutting-edge Low-Latency AI, and Core AI that integrates TenyxChat—a developer-friendly, open-source multi-Large Language Models (LLMs) architecture from HuggingFace, optimized for voice applications.

Tenyx’s industry-specific voice Agents will enable enterprises to deliver seamless, scalable, and personalized customer interactions.

Tenyx’s voice platform addresses key frustrations of traditional interactive voice response (IVR) systems, such as poor intent recognition and lack of dynamic scaling, while also overcoming the challenge of optimizing AI for voice rather than text. Their solutions span industries like e-commerce, hospitality, consumer services, and market research.

The Tenyx team—led by co-founders Itamar Arel (CEO) and Adam Earle (CTO)—will join Salesforce to further advance the innovation within Salesforce’s Service Cloud. By leveraging Tenyx’s voice AI capabilities like endpointing, dynamic scaling, and turn-based conversational AI, Salesforce aims to redefine enterprise customer service, allowing businesses to enhance customer satisfaction, improve agent efficiency, and boost revenue through real-time, low-latency voice interactions.

Salesforce’s acquisition of Tenyx is expected to close by the end of Q3 of Salesforce’s fiscal year 2025, on October 31, 2024, subject to customary closing conditions.

Tenyx held a StartupHub.ai AI technology score of 100, backed by a team of six AI professionals, with key research published on LLMs published on Arxiv, including studies on forgetting in fine-tuning LLMs and geometric characterizations of LLMs. The startup also released an open-source, fine-tuned LLM, Tenyx 70B. It was built from Meta’s open-source Llama-3 language model, and outperformed OpenAI’s GPT-4 in certain domains at the time of publishing.

Founded in 2022, Tenyx had raised $15 million from investors including StageOne Ventures, Pathbreaker Ventures, Morado Ventures, Point72 Ventures, Cota Capital, and AME Cloud Ventures.

The acquisition marks StageOne Ventures' continued focus on AI voice technology. The firm has clear track record in the subsector, having previously invested in Apprente, which was acquired by McDonald’s to power AI-driven voice technology in its drive-thrus. A similar StageOne portfolio startup is Second Nature, using conversational AI to provide sales teams with life-like training simulations.