In a move that signals a tectonic shift in the development of artificial intelligence, the AI coding startup Cognition has announced its acquisition of The Interaction Company of California, the creators of Poke. The deal, which industry insiders value in the "low nine figures," marks a strategic pivot toward prioritizing the "personality" and "interaction model" of AI agents as much as their technical performance. By bringing Poke—the AI assistant famous for its conversational, friend-like interface—under the Cognition umbrella, the company aims to fundamentally transform its flagship product, Devin, from a mere coding tool into a persistent, approachable, and highly capable digital colleague.
The Genesis of the Deal: Merging Tech with Tone
The acquisition brings together two of the most talked-about entities in the current AI landscape. Cognition, founded by Scott Wu and Walden Yan, has gained significant traction with its "Devin" coding agent, which is designed to handle complex software engineering tasks autonomously. The Interaction Company, co-founded by Marvin von Hagen, made waves with the launch of Poke in March 2026.
Unlike traditional, rigid AI assistants that function like command-line utilities, Poke was built to mirror human social interaction. It utilizes messaging platforms like iMessage, SMS, Telegram, and WhatsApp to engage users in casual, slang-heavy, and humorous dialogue. This "human-in-the-loop" approach to AI interaction is precisely what attracted the leadership at Cognition. For Wu and Yan, who were early angel investors in The Interaction Company, this acquisition is not merely a purchase of intellectual property—it is the integration of a specific cultural philosophy into the world of high-stakes software engineering.
Chronology: From Messaging App to Enterprise Asset
The rise of Poke was meteoric, driven by a user base that craved a more natural way to interact with their AI tools.
- March 2026: The Interaction Company launches Poke, an AI agent accessible through standard messaging apps. It quickly finds a niche in travel planning, financial tracking, and scheduling.
- June 2026: Poke achieves a significant industry milestone, becoming the first AI agent approved to operate on Apple’s "Messages for Business" platform, signaling institutional trust in the startup’s safety and utility standards.
- Mid-2026: User engagement spikes, with over 100 million messages exchanged in just three months. Despite high usage, the operational costs of maintaining such a high-fidelity, real-time agent become a hurdle for independent profitability.
- Late 2026: Cognition moves to acquire the company. The deal is finalized, bringing the team and the technology into the Cognition ecosystem.
- 2027 and Beyond: The integration phase begins, with plans to layer Poke’s interaction architecture over Cognition’s powerful SWE-1.7 software engineering models.
Supporting Data: Why Personality Matters
The success of Poke was never just about its ability to answer questions; it was about the delivery. According to data released by the companies, Poke garnered hundreds of thousands of active users who treated the AI less like a search engine and more like a personal assistant or peer.
However, the reality of running a high-frequency, low-latency AI service is capital-intensive. Marvin von Hagen candidly admitted that while the engagement numbers were staggering—100 million messages in a single quarter—the underlying infrastructure costs made sustaining the business model as an independent entity difficult. The acquisition provides the necessary financial backbone from Cognition to scale these interactions without the constant pressure of immediate profitability, allowing the development team to focus on the qualitative "feel" of the agent.
Official Perspectives: Building the "Colleague"
The vision for this partnership is rooted in the belief that the future of work is collaborative, not just automated. In a recent blog post, Cognition co-founder Scott Wu summarized the synergy: "The Interaction team has built an agent that people love: it’s proactive, it knows you, and it’s fun to talk to. That’s exactly how working with Devin should feel, and now we get to build it together."
Marvin von Hagen, speaking with TechCrunch, underscored the psychological importance of this shift. "You probably prefer it if you have co-workers that have personality, rather than if you have co-workers that are just robots," von Hagen said. "When you have co-workers that are also software engineers, they can also make a joke. And you’ll find it funny and enjoyable."
This transition marks a departure from the "black box" era of AI. The goal is to create a sense of continuity. Currently, AI agents often suffer from "session amnesia"—they forget the nuances of previous interactions. By leveraging Poke’s architecture, Cognition aims to create a "persistent co-worker" that remembers past projects, understands the user’s specific preferences, and maintains a stable personality throughout long-term engineering tasks.
Implications for the Future of AI Agents
The acquisition of The Interaction Company suggests that we are entering a new phase of the AI gold rush: the battle for the "User Experience (UX) of Intelligence."
The Move Toward Orchestration
One of the most promising aspects of this merger is the potential for Poke to act as a manager for Devin. Currently, Devin is limited by its scope; it typically handles one pull request at a time. Von Hagen envisions a future where Poke acts as an orchestrator, managing multiple Devin sessions and keeping track of complex, multi-layered projects across different timeframes. This would effectively turn the AI into a project manager that understands the "big picture" rather than just the task at hand.
Apple and the Business Ecosystem
Poke’s integration with Apple’s Messages for Business remains a key strategic asset. By maintaining its presence on the platform, Cognition is positioned to reach users where they are most comfortable. This is a subtle but powerful play: by embedding a high-level coding agent into a platform as ubiquitous as iMessage, Cognition is lowering the barrier to entry for complex software engineering.
The Shift in Competitive Advantage
For the broader tech industry, this deal serves as a warning: technical accuracy is becoming a commodity. As models like OpenAI’s o1 or Anthropic’s Claude achieve higher reasoning capabilities, the differentiator will no longer be "can the AI write the code?" but rather "how does it feel to work with the AI?"
Companies that ignore the "soft skills" of AI—personality, humor, proactive communication, and memory—risk building tools that are efficient but ultimately cold and difficult to integrate into creative or collaborative workflows.
Conclusion: A New Era of Collaboration
The road ahead for the combined team involves a period of integration, with plans to implement Cognition’s latest SWE-1.7 software engineering model into the Poke framework starting in the coming year. Users should not expect immediate changes, as the companies plan to maintain the status quo through the end of the year to ensure stability.
However, the long-term roadmap is clear. By synthesizing the raw power of Cognition’s engineering models with the conversational, empathetic interface of Poke, the companies are attempting to build the "ideal colleague." If successful, this merger will redefine the standard for AI agents, moving the industry away from the era of "chatbots" and into the era of "digital teammates."
As the lines between human and machine collaboration continue to blur, the success of this acquisition will be measured not by the complexity of the code produced, but by the rapport built between the human developer and their silicon-based partner. In the competitive landscape of 2027, the AI that gets the joke might just be the one that wins the market.
