Poke, a startup backed by Spark Capital and General Catalyst, is positioning its AI agent as accessible through iMessage, SMS, Telegram, and, in some markets, WhatsApp. The company launched publicly in March and enables users to perform everyday tasks and automations through text-based interactions, without requiring downloads, signups, or technical setup. The approach reflects a shift in AI toward “agentic” systems that can take actions on a user’s behalf rather than only answer questions.
AI agents delivered through familiar messaging
According to TechCrunch, Poke’s core product is an AI agent accessible via messaging platforms rather than a dedicated app interface. The service functions as a personal assistant that can “take action on their behalf” through iMessage, SMS, Telegram, and—depending on market—WhatsApp. This design choice affects adoption: if user interaction occurs within a phone’s existing messaging interface, the barrier to first use is lower than for tools requiring new software installation or separate workflows.
TechCrunch distinguishes Poke’s use case from general-purpose chatbots like ChatGPT or Claude. While users may consult those systems for questions and research, Poke targets users who want to “get something done quickly” or automate tasks. In other words, Poke addresses a different usage pattern than chat-only systems—one focused on task execution and ongoing reminders.
According to TechCrunch, Poke can assist with daily planning, calendar management, health and fitness tracking, smart home control, photo editing, and other everyday needs. The common thread is that tasks are triggered through text.
From reminders to automations: plain text as the control layer
TechCrunch provides concrete examples of Poke’s capabilities. Users can ask Poke to alert them to specific emails (such as messages from family or a boss), remind them in the morning to take an umbrella, track health and fitness goals, notify them about sports scores, send daily medication reminders, or provide news updates.
A key feature is how Poke handles automation creation. According to TechCrunch, users can “write their own automations in plain text and then share them with friends.” This workflow allows user instructions—rather than a complex configuration interface—to become the specification for what the agent should do.
In the broader landscape of AI agents, this approach differs from many systems that require integration steps, permissions, or specialized setup. TechCrunch’s description suggests that Poke aims to collapse those steps into a conversational or message-based interface. The user-facing design described indicates a shift toward “automation as message,” where the same channel used for communication becomes the channel for task orchestration.
Public launch in March and funding details
TechCrunch reports that Poke launched publicly in March, making the agent available to consumers. The startup is a 10-person company backed by Spark Capital, General Catalyst, and other angel investors. According to the source, Poke “has more recently added another $10 million to its coffers, on top of last year’s $15 million seed round.” The company is valued at $300 million, post-money.
This funding level places Poke within a category of well-funded AI startups competing in the “agentic AI” systems space. TechCrunch also references a video and social post from Poke (@interaction) dated March 19, 2026, which mentions “Introducing Poke Recipes,” “Create a Recipe in 10 seconds,” “Earn on Poke,” and “Build with npx.” The source does not provide full technical details about these features.
Message-based agents and market demand
TechCrunch notes that “demand for agentic AI systems is spiking,” positioning Poke within a broader market trend. The source mentions OpenAI in relation to this demand, though the complete context is not fully detailed in the provided excerpt.
The core product design described by TechCrunch—agents accessible through standard messaging—offers insight into how AI functionality may be packaged for everyday users. If agents are delivered through SMS-like channels, the interaction surface becomes the messaging workflow users already have: send a text, receive an action or update, and continue the conversation as needed.
For developers and product teams, a message-based approach could influence how integrations are built. TechCrunch describes Poke’s capabilities across smart home devices and photo editing, though it does not specify underlying integration methods. The described capabilities suggest the agent must translate plain-text instructions into actions across multiple domains—calendar, fitness tracking, photo editing, and smart home control—within the constraints of a chat-like interaction model.
Observers may watch for whether other AI agent products adopt similar “no app, no signup” access models, or whether messaging becomes a standard interface for agent execution. TechCrunch’s emphasis on eliminating setup friction suggests that competition in this space may increasingly focus on how quickly an agent can turn user intent into completed tasks, rather than model capability alone.
What to look for next
Based on the TechCrunch report, Poke’s near-term differentiators include its messaging accessibility, the range of everyday tasks it supports, and the ability for users to author automations in plain text and share them. The funding and valuation snapshot indicates investor interest in agentic systems delivered through consumer-friendly interfaces.
For technologists evaluating the approach, a key question is how Poke handles reliable execution—especially for reminders, calendar actions, and smart home control—when the interface is a short message. TechCrunch does not provide implementation details, but the product’s stated capabilities outline the kinds of integrations that would need to be robust for the system to function as an assistant that “takes action” on a user’s behalf.
Source: TechCrunch