Sundae Personal AI Agent Launches via Text Message

Sundae Bar AI PLC has entered the consumer AI market with the beta launch of the Sundae personal AI agent, a product that lets people delegate everyday tasks through ordinary text messages. According to the company's announcement on October 21, 2026, the new agent works through iMessage, RCS on Android, and WhatsApp, with no application to download and no new interface to learn.

The pitch is deliberately simple: instead of opening a chatbot and writing prompts, users send a message the way they would text a colleague. The Sundae personal AI agent then takes the task from there, using its own computer, browser, and persistent memory to research a question, create content, build a small digital project, or keep watch over information in the background.

That last point matters. The agent is designed to retain context between conversations and to notify users proactively when something it is monitoring changes. According to the announcement, the goal is to let people hand over ongoing work rather than repeat the same request every time they need an update.

How the Sundae personal AI agent works

Under the hood, the Sundae personal AI agent combines several capabilities that have become standard in the agent world: web browsing, tool use, and memory that persists across sessions. What sets the product apart is the surface it presents. By living inside messaging apps people already use, it removes the friction of adopting yet another AI tool.

The company reports that the beta drew 150 sign-ups almost immediately after launch, enough that it increased testing capacity and then opened a waitlist as demand continued. The product remains in beta while the company studies how people use it and what commercial opportunities it opens up.

For the wider agent ecosystem, the launch is another sign that personal agents are moving from developer demos into consumer products. Recent coverage of voice agents that run locally, such as Block's Dessa voice model, shows the same pattern: capable agents packaged so ordinary users can adopt them without technical setup.

Built on an agent-testing engine

Sundae did not start as a consumer idea. According to the company, it grew out of infrastructure built to evaluate thousands of AI Skills and Agent submissions for its commercial platform. That infrastructure gave agents a computer, a browser, and tools, then measured how well they completed tasks.

The testing ground is Subnet 121, the company's AI development network on Bittensor, where developers around the world compete to build and improve AI capabilities. Since that network launched, the company says it has run 30 development challenges and received nearly 7,000 competing submissions.

That history gives the Sundae personal AI agent a foundation most startups lack. The same systems that graded thousands of agents now power one agent aimed at individuals. The company frames this as efficient reuse: one technology investment supporting a second product without building a separate platform from scratch.

One foundation, two markets

Sundae Bar's core business remains enterprise AI. Through a subscription platform, a marketplace, and bespoke deployments, it gives businesses access to a library of more than 700 AI Skills and Agents covering marketing, finance, research, and workflow automation.

The consumer launch does not change that strategy, according to the announcement. Instead, it tests whether the same technology can serve individuals directly, while the enterprise business keeps growing. Chief executive Jill Kenney described the move as bringing to individuals a capability previously aimed at businesses, built on technology developed to test thousands of agents.

The experiment also fits a broader shift in enterprise software, where agents are gaining their own workplace identities. Coverage of Gemini Launchpads for federal agencies shows agents moving from pilots into mission operations, a sign that agent infrastructure is maturing on both sides of the consumer-enterprise divide.

Whether the Sundae personal AI agent finds a lasting consumer audience will depend on the beta. Messaging-based agents face real questions about trust, privacy, and how much autonomy users will grant a helper that acts in the background. The company says it is focused on testing functionality and understanding user behaviour before making bigger commercial bets.

For now, the Sundae personal AI agent is a small but telling launch: an agent company turning its own testing tools into a product, and betting that the easiest way to reach consumers is to meet them where they already type. More information is available at sundae.md, and the full announcement was carried by TradingView and ADVFN.