A small Palo Alto startup is trying to remove one of the biggest frictions in artificial intelligence adoption: the need to open a separate app. Its solution is simple on the surface and ambitious underneath. Turn an AI assistant into a contact that lives inside your messaging inbox.
That is the premise behind Poke, an emerging AI product that operates entirely through text messages. Instead of downloading an app or navigating a dashboard, users interact with the assistant the same way they would text a friend.
Poke does not begin with installation. Users visit its website, enter a phone number, and the assistant begins texting them directly. From that point forward, all interactions happen inside existing messaging platforms such as iMessage, SMS, and RCS.
This approach removes onboarding complexity, which has been one of the biggest barriers to consumer AI adoption. There are no plugins, no integrations to configure manually, and no separate interface to learn. Instead, users issue commands in plain language, such as setting reminders, tracking workouts, or managing emails.
The design reflects a broader shift in AI products. Rather than building new environments, companies are starting to embed intelligence into tools people already use daily. Messaging, in this case, becomes the interface layer for automation.
Unlike traditional AI assistants that focus on answering questions, Poke is positioned as an action-driven system. It does not just respond to prompts but executes tasks across different services.
Current capabilities include managing calendars, setting both time-based and context-based reminders, monitoring inbox activity, and summarizing workouts through integrations like Strava. It can also control smart home devices, provide daily news updates, and handle lightweight tasks such as travel suggestions or generating exercises for students.
A notable feature is its “recipes” system, which allows users to install pre-built automations written in natural language. These recipes cover categories ranging from productivity and finance to health and developer workflows. Users can also create and share their own automations, turning the platform into a distributed library of text-based workflows.
For more technical users, Poke extends into developer tooling. It integrates with platforms like GitHub, Supabase, and Vercel, allowing parts of development and analytics workflows to be triggered through simple messages.
Poke is developed by The Interaction Company, a California-based startup founded by Marvin von Hagen and Felix Schlegel. The team previously worked on an AI email assistant, but user behavior revealed a broader opportunity.
Early adopters were pushing the product beyond email into general task management. That insight led the team to pivot toward a more expansive assistant controlled entirely through text.
The company has since raised $25 million in funding, including a $15 million seed round in 2025 and an additional $10 million in 2026. Investors include Spark Capital and General Catalyst, with the startup reportedly reaching a $300 million valuation despite operating with a team of around ten people.
During its closed beta phase, Poke attracted a concentrated group of early users, including founders, investors, and operators from major technology companies. These users exchanged hundreds of thousands of messages with the assistant each month.
Across the first few thousand users, the company reported over 750,000 total messages and described retention as nearly perfect. While these numbers come from a limited and highly technical audience, they suggest that the messaging-first model may resonate with users who are already accustomed to automation tools.
The real test, however, will come when the product expands beyond early adopters and into broader consumer use.
Poke does not rely on a single AI model. Instead, it routes requests across multiple models depending on the task, selecting between proprietary systems from major AI providers and open-source alternatives.
This multi-model approach allows the system to balance cost, performance, and response quality. It also avoids dependency on a single vendor, which has become a strategic concern for many AI startups.
On the infrastructure side, Poke uses a messaging platform called Linq to embed itself into existing communication channels. This is what allows it to function as a “contact” rather than an application, handling message delivery and routing without requiring a separate interface.
The company’s long-term vision extends beyond reactive commands. Its roadmap focuses on making the assistant more proactive.
Rather than waiting for instructions, Poke aims to detect patterns and surface relevant actions automatically. For example, it could remind users about overdue invoices, suggest follow-ups on unanswered emails, or highlight changes in schedules without being prompted.
This shift from reactive to proactive behavior is central to the broader evolution of AI agents. The challenge will be balancing usefulness with intrusiveness, especially as the assistant gains access to more personal data.
Poke enters a rapidly expanding category of AI agents, many of which remain heavily technical and require configuration through dashboards, APIs, or code.
Its core differentiation is simplicity. By using messaging as the interface, it lowers the barrier to entry for non-technical users. Instead of setting up workflows manually, users describe what they want in plain language and let the system handle execution.
This positions Poke as a consumer-facing alternative to more complex agent frameworks, which are typically designed for developers or enterprise users.
The approach also aligns with a broader industry trend. AI is increasingly being embedded into existing tools rather than introduced as standalone platforms. Messaging, in this case, becomes the natural layer where automation meets everyday behavior.
Despite early enthusiasm, several questions remain unresolved.
Most current coverage focuses on features and design rather than long-term reliability. As Poke expands, it will need to demonstrate consistent performance across more complex workflows, especially when handling tasks that involve external systems like email, calendars, and financial data.
Error handling is another area to watch. When an AI assistant takes action on behalf of a user, even small mistakes can have larger consequences. The balance between automation and control will likely define how widely such tools are adopted.
Data security will also become a critical concern. As the assistant gains access to sensitive information across multiple services, trust will depend on how well the system protects and manages that data.
Poke represents a shift in how AI assistants are delivered, not just what they do.
By embedding itself into messaging platforms, it removes friction and makes AI interaction feel natural. It simplifies automation to the point where using it feels less like operating software and more like sending a text.
Whether that simplicity can scale into reliability remains to be seen. But as of now, Poke stands as one of the clearest early examples of a consumer-ready, text-native AI agent, signaling where the next phase of AI adoption may be headed.
The National Security Agency is reportedly using Anthropic’s restricte...
If editing videos had a gym membership… most people would cancel in we...
Every startup needs a brand identity before it needs almost anything e...
Whether its the logo, Facebook post, website, product photos, or video...
Most business owners check their Google Business Profile to update hou...
Discussion