The Pulse
SoundHound Puts OASYS Voice Agents on Edge Devices
SoundHound announced OASYS Edge, a system for deploying voice agents on vehicles and smart devices as well as in the cloud. The company says local processing can keep selected interactions working without a network connection, with deployme
AI.info Team ·
One OASYS voice agent, three deployment options: cloud, device, or a mix of both. SoundHound AI announced OASYS Edge on September 24, pitching the system for vehicles and smart devices that need conversational AI even when a network connection drops.
The company says developers can build an agent once on its OASYS platform, then choose where it runs based on the product and use case. Local processing is intended to keep some conversations and actions on the device, while tasks that need an online service can still use the cloud. SoundHound says deployment is scheduled for late 2026; it says demonstrations are available now and plans to feature the technology at CES 2027.
Local agents for requests that span multiple systems
SoundHound describes OASYS Edge as more than an offline version of a conventional voice assistant. Its announcement says the system is designed to interpret ambiguous requests and coordinate actions across multiple device functions using purpose-built language models. That could mean responding to a driver’s broader need rather than matching a request to one predefined command.
One example in the release involves a driver with poor signal asking for a route to Los Angeles with two tourist stops and descriptions of each. SoundHound says OASYS Edge could use a local knowledge base to plan the stops and narrate them without a cloud connection. The release also describes a robot that handles query detection and everyday conversational responses on-device instead of continuously sending audio to the cloud.
SoundHound links local processing to privacy and cost
Running interactions locally could reduce the amount of user data sent to external servers and the need for repeated cloud requests, the company says. SoundHound presents those as potential privacy and operating-cost benefits for manufacturers, alongside faster responses and continued operation in places such as tunnels or remote areas. The announcement says interactions, voice processing and telemetry can optionally remain on-device; it does not say every task will be handled locally.
“This allows them to provide uninterrupted AI assistance to end users without network overhead or cloud processing,” said James Hom, SoundHound AI’s chief product officer and co-founder, describing the flexibility of the deployment model.
Deployment is planned for late 2026
The announcement does not name a launch customer, supported chipsets, minimum hardware requirements or pricing. It says the system will offer customizable model sizes and configurations, but does not provide benchmarks for response time, offline accuracy or the range of tasks that will work without a connection.
Those details will determine how far OASYS Edge can move beyond the company’s demonstrations into vehicles and consumer devices. For now, SoundHound’s concrete commitment is a late-2026 deployment schedule, with demonstrations available ahead of a planned CES 2027 showcase. The company’s announcement sets out the proposed architecture, but leaves the hardware and performance specifics for a later stage.