AI agents are moving from developer tools into daily workflows, and Poke is positioning itself as a way for consumers to access that capability through messaging apps. According to TechCrunch, the startup launched its AI agent publicly in March, and it can be reached via iMessage, SMS, Telegram, and, in some markets, WhatsApp. The pitch: make “doing” and “automating” tasks as simple as sending a text.
Messaging as the interface for agent actions
Poke frames the AI agent as a tool users turn to when they want to complete tasks, rather than as a conversational chatbot. The agent functions as a personal assistant that can take action on a user’s behalf through a familiar interface—text messages—without complex setup, apps, or technical know-how.
Poke can support everyday needs such as daily planning, managing your calendar, tracking health and fitness, controlling a smart home, and editing photos. This usage pattern differs from general-purpose chatbots: users might use systems like ChatGPT or Claude for questions or research, while turning to Poke when the goal is to complete tasks quickly or automate them to save time.
What Poke can do through text
Poke can handle a range of routine requests. Users can ask it to alert them to specific emails (including messages from family or a boss) or to remind them in the morning if they need to take an umbrella. The agent can also help users track health and fitness goals and notify them of game scores.
Beyond reminders and updates, Poke can send daily medication reminders and catch users up on the day’s news. A key feature is that users can write their own automations in plain text and then share them with friends, lowering the barrier to creating agent workflows. These user-defined automations are called “Poke Recipes,” and they can be created quickly without requiring technical expertise.
Accessibility and automation workflows
Poke is designed to be accessible without friction. The product emphasizes “No download, no signup,” allowing users to start using it by simply texting. This approach means users can access the agent through channels they already use for communication, rather than installing a dedicated app or registering through a separate interface.
The technology challenge for Poke centers on reliably turning text messages into actions—calendar updates, smart home control, photo edits, and notification-style reminders. The product targets these “actionable” categories, enabling users to automate routine tasks without manual intervention.
Funding and market timing
Poke is backed by Spark Capital, General Catalyst, and other angel investors. The 10-person startup recently added $10 million to its coffers on top of a $15 million seed round from the previous year. The company is now valued at $300 million post-money.
Poke’s launch arrives as demand for agentic AI systems is increasing. The startup is one of several companies attempting to operationalize AI beyond conversational chat, positioning agents as tools for task completion and automation rather than as general-purpose assistants.
Why this matters for product design
Poke’s approach suggests a direction for consumer AI: if agents become widely useful when triggered through everyday channels, messaging-based interfaces could become a common pattern for agent tools. By routing agent capabilities through iMessage, SMS, Telegram, and WhatsApp, Poke aligns itself with existing user habits rather than asking users to adopt a new app or workflow.
The examples—email alerts, umbrella reminders, medication notifications, daily news catch-ups, and smart home control—represent tasks that can be represented as scheduled or event-driven actions. User-defined automations written in plain text could serve as a bridge between natural-language instructions and repeatable workflows that users can reuse and share.
Whether this messaging-first model becomes standard will likely depend on how consistently the agent executes actions and how safely it handles requests, particularly when managing calendars, health tracking, or device control.
Source: TechCrunch