While the tech world buzzes with complex “agentic AI” systems—tools that don’t just talk, but actually do things—most people find them intimidating. Setting up these systems often requires navigating technical terminals, managing software dependencies, and troubleshooting errors.
Poke is aiming to change that by stripping away the complexity. Instead of a complicated interface, Poke allows users to interact with a powerful AI agent through the apps they already use every day: iMessage, SMS, Telegram, and WhatsApp.
From Email Assistant to Personal Life Manager
The startup behind Poke, the Palo Alto-based The Interaction Company of California, didn’t start with a general-purpose assistant. Originally, they built an AI specifically for managing email. However, they quickly noticed a trend: users weren’t just asking about emails; they were asking for medication reminders, sports scores, and weather updates.
This insight led to a pivot. Rather than a niche tool, Poke evolved into a versatile personal assistant designed to be proactive and “human” in its interactions.
How It Works: The Power of “Recipes”
The core of Poke’s utility lies in its “recipes” —pre-made automations that connect the AI to your existing digital life. Instead of writing code, you simply activate a recipe to link Poke to your favorite services.
- Productivity & Work: Integrations with Gmail, Google Calendar, Outlook, Notion, and Linear.
- Health & Wellness: Connection to Strava, Oura, Fitbit, and Withings.
- Smart Home: Control over devices like Philips Hue and Sonos.
- Developer Tools: For technical users, Poke integrates with GitHub, Vercel, Supabase, and more.
By using these recipes, Poke acts as a bridge between your text messages and your digital ecosystem. You can ask it to alert you to specific emails, track fitness goals, or even remind you to bring an umbrella based on the morning forecast.
A Flexible, Model-Agnostic Approach
One of Poke’s strategic advantages is its independence. While giants like Meta or OpenAI are locked into their own proprietary models (Meta AI can only use Meta models, for example), Poke is “model-agnostic.”
Under the hood, Poke selects the best AI model for the specific task at hand, whether it is a massive model from a major provider or a specialized open-source version. This flexibility allows it to remain highly efficient and versatile.
Business and Growth
The startup is backed by heavyweights like Spark Capital and General Catalyst, with a recent $10 million infusion bringing its valuation to $300 million. Its investor list reads like a “who’s who” of Silicon Valley, including the founders of Stripe, PayPal, and Dropbox.
“We really don’t want to make money, but we really want to grow. We want to build a product for a billion people,” says co-founder Marvin von Hagen.
Currently, Poke’s growth strategy focuses on user acquisition rather than immediate profitability. Their pricing model is uniquely dynamic:
– Free Tier: For basic tasks that don’t require real-time data.
– Usage-Based Pricing: For tasks requiring “real-time inference” (like monitoring live flight updates or incoming emails), the cost is determined by the complexity and data requirements of the task.
Challenges and the Road Ahead
The path to global adoption isn’t without hurdles. Meta has restricted general-purpose chatbots on WhatsApp, a move currently under scrutiny by regulators in the EU, Italy, and Brazil. Poke is navigating these regulatory waters while attempting to expand its reach.
To build a community, Poke is also incentivizing creators. Developers and influencers can build their own custom recipes and earn a small commission (between 10 cents and $1) for every user who signs up via their specific automation.
Conclusion
By moving AI agents from the technical terminal to the text message inbox, Poke is attempting to democratize automation. If successful, the “agentic” future won’t be something you program, but something you simply text.
