AI agent poke lets you set up automations by texting

At a glance:

• Poke is a text‑based AI assistant that runs on iMessage, SMS, Telegram and limited WhatsApp support.
• Users create and share plain‑text "recipes" to automate tasks like calendar management, health tracking and smart‑home control.
• Backed by Spark Capital and General Catalyst, the 10‑person startup has raised $25 million and is valued at $300 million.

how poke works and why it matters

Poke positions itself as a consumer‑friendly alternative to heavyweight agentic AI platforms such as OpenClaw. Rather than requiring a terminal install or a dedicated app, users simply visit poke.com, enter a phone number and start chatting. The assistant lives inside existing messaging channels, leveraging a service called Linq to embed AI capabilities in iMessage, SMS and Telegram. This low‑friction onboarding is designed for people who want to get things done quickly without learning new interfaces.

Under the hood, Poke routes each request to the most suitable large‑language model, whether that’s an OpenAI, Anthropic or open‑source offering. By remaining model‑agnostic, the startup claims a strategic advantage over rivals that are locked into a single provider’s ecosystem. The flexibility also lets Poke scale its pricing model based on the computational cost of each automation – real‑time inference (e.g., live flight check‑ins) is billed, while static queries can remain free.

the recipe ecosystem: plug‑and‑play automations

At launch, Poke shipped a library of pre‑built “recipes” that automate common workflows. Categories span health (Strava, Fitbit, Withings), productivity (Gmail, Google Calendar, Outlook, Notion), smart‑home devices (Philips Hue, Sonos) and developer tools (GitHub, Vercel, Supabase). Installing a recipe is a single click followed by an OAuth‑style permission grant, after which users can trigger the automation with plain‑text commands such as “Hey Poke, remind me to take my meds at 8 am.”

The platform encourages community contribution: creators earn between $0.10 and $1.00 per new user who adopts a recipe they authored. In the weeks following the public launch, thousands of user‑generated recipes were added, and Poke plans to surface them in a searchable directory. This crowdsourced approach mirrors the early growth tactics of app stores, but with the added benefit of instant, text‑based activation.

security, privacy and regulatory hurdles

Poke’s security model is multi‑layered. Regular penetration testing, token‑level permission scoping and a policy that prevents internal staff from viewing token contents by default are all part of the design. Users must explicitly enable logging or analytics sharing, which limits data exposure.

Nevertheless, the service faces regulatory friction. Meta’s decision to charge high fees for WhatsApp chatbot access has blocked full WhatsApp integration, prompting antitrust probes in the EU, Italy and Brazil. Poke’s founder Marvin von Hagen describes the fees as “malicious compliance” and expects pressure to ease the cost, which would unlock a massive user base in regions where WhatsApp dominates messaging.

market positioning and growth outlook

The timing of Poke’s debut coincides with a surge in interest for agentic AI, highlighted by OpenAI’s acquisition of OpenClaw’s creator and Nvidia’s push for enterprise‑grade alternatives. While those solutions target developers and large organizations, Poke aims at the mainstream consumer market, emphasizing ease of use over deep system access.

Financially, the startup has raised a total of $25 million—$15 million seed in 2024 and a $10 million extension in 2026—backed by Spark Capital, General Catalyst and a roster of high‑profile angels including the Collison brothers (Stripe), John Collison, and OpenAI’s Joanne Jang. Valued at $300 million post‑money, the company is prioritizing user acquisition over profitability, stating that monetization is “secondary” to building a product for a billion people.

If Poke can sustain rapid recipe growth, secure broader WhatsApp access and keep operational costs low, it could become a de‑facto standard for personal AI automation, challenging the more technical, install‑heavy alternatives that dominate the enterprise space.

future developments and user experience expectations

Looking ahead, Poke plans to expand its recipe catalog, improve real‑time inference pricing transparency, and deepen integrations with emerging APIs. The team also hinted at a creator‑focused marketplace where developers can sell premium automations, potentially adding a new revenue stream.

From a user perspective, the most compelling promise is the ability to treat a chatbot like a personal concierge that acts on your behalf without ever leaving the messaging app you already use. As AI models become cheaper and more capable, the line between “chatting” and “executing” will blur, and Poke’s text‑first approach may set the template for the next generation of consumer AI agents.