Cognition Snaps Up Friendly AI Texting App Poke to Give Coding Agent Devin a Human Touch
Key takeaways
- Cognition acquired Poke's parent company in a low nine-figure deal to bring its conversational, friend-like personality to the Devin coding agent.
- Poke logged over 100 million messages in three months across hundreds of thousands of users, but struggled with high operating costs before the acquisition.
- Future plans include using Cognition's SWE-1.7 model to enhance Poke and potentially letting Poke orchestrate multiple simultaneous Devin sessions with persistent memory.
Cognition, the startup behind the autonomous AI coding agent Devin, has acquired The Interaction Company of California — the team behind Poke, an AI assistant designed to communicate with users through familiar messaging platforms like iMessage, SMS, Telegram, and WhatsApp. The deal, confirmed by Interaction Company co-founder Marvin von Hagen in an interview with TechCrunch, values the startup in the 'low nine figures,' marking one of the more notable AI acquisition price points of the year. Cognition co-founders Scott Wu and Walden Yan were previously angel investors in the company, giving the deal a personal dimension beyond simple market competition.
Poke launched in March 2026 and quickly distinguished itself from the crowded AI assistant space by leaning into personality rather than raw capability. Rather than responding with stiff, formal outputs, Poke engages users with humor, slang, and a proactive style that mimics how a friend might reply to a text message. People have been using the assistant across a surprisingly wide range of tasks — from travel planning and health queries to finance, scheduling, and education — with productivity staples like email management and reminders topping the usage charts.
The scale of Poke's engagement is difficult to ignore: users exchanged over 100 million messages on the platform in just the past three months, across hundreds of thousands of active users. Despite that traction, von Hagen acknowledged that Poke had been costly to operate, making profitability elusive. The acquisition by a better-resourced Cognition solves an existential infrastructure problem while simultaneously giving Cognition something money alone can't easily buy: a proven approach to human-AI rapport.
For Cognition, the strategic logic is clear. Devin has made headlines as a capable autonomous software engineer, but capability alone may not be enough to retain users in an increasingly competitive AI agent market. By folding Poke's interaction philosophy into Devin, Cognition hopes to make its coding agent feel less like a tool executing commands and more like a genuine collaborative colleague. Scott Wu captured this in a blog post, writing that Poke is 'proactive, it knows you, and it's fun to talk to — that's exactly how working with Devin should feel.'
Looking ahead, Poke users won't see immediate changes through the end of the year, and the product will remain active on Apple's Messages for Business platform — where it became the first AI agent approved to operate — through that period. In 2026, however, the teams plan to experiment with deeper integration, including routing some of Poke's tasks through Cognition's latest SWE-1.7 software engineering model. Von Hagen also floated the intriguing possibility of Poke acting as an orchestration layer across multiple simultaneous Devin sessions, effectively allowing the friendly assistant to manage persistent memory and multi-task in ways that Devin currently cannot handle alone.
The bigger picture
This acquisition is a clear signal that the AI industry is entering a second phase of competition — one where the interface layer matters as much as model capability. For the past few years, companies poured resources into making their underlying models smarter, faster, and cheaper. Cognition's purchase of Poke suggests the frontier is shifting toward something harder to benchmark: whether users actually enjoy interacting with their AI. That's a genuinely difficult problem to solve with engineering alone, and buying a team that has already cracked it — proven by 100 million messages in three months — may be far more efficient than building from scratch.
There's also a meaningful competitive signal here for the broader coding agent market. Devin launched to enormous fanfare but has faced skepticism about whether its real-world utility justifies the hype. Adding a personality layer could be the differentiator that makes Devin sticky for developer teams rather than a tool they try once and abandon. Microsoft's GitHub Copilot and Google's various coding assistants are formidable rivals, but neither has made warmth or conversational fluency a headline feature. If Cognition executes well, it could carve out a genuinely distinct positioning in a market that currently competes almost entirely on benchmark scores.
Investors and founders in the AI space should watch how this integration unfolds over the next 12 months. The idea of Poke orchestrating multiple Devin sessions — maintaining memory across tasks and managing parallel workstreams — is ambitious and, if it works, could be transformative for enterprise developer workflows. The risk is that blending two products with distinct identities and user bases produces something incoherent rather than complementary. Cognition's angel investor history with the Poke team may help cultural alignment, but product mergers in fast-moving AI startups rarely go smoothly. The outcome here could either validate personality-first AI strategy or serve as a cautionary tale about acquisition-driven product thinking.
We're covering this deal at LagPing because it gets at something we think about a lot: the gap between what AI can do and what AI feels like to use. Most benchmarks measure capability, but almost none measure whether a tool makes you want to keep coming back. Poke is interesting precisely because it built a following not on being the smartest assistant, but on being the most pleasant to talk to — and that's a genuinely different design philosophy. As AI agents become a bigger part of how developers and everyday users get things done, the question of personality and rapport is going to matter more, not less. We also think the Devin angle here is worth watching closely given how much attention autonomous coding agents have attracted from both enterprise buyers and the developer community. This isn't just a startup acquisition story — it's a preview of where the AI assistant market may be heading, and we want our readers to understand the stakes before it becomes obvious.
As an Amazon Associate, LagPing earns from qualifying purchases. Product links are affiliate links.
You might also like
Wikipedia's Parent Organization Faces Growing Worker Rebellion as US Staff Demand Recognition
Jul 29
GTA6 Studio's Worker Union Hits Record Membership, Threatening Forced Recognition Battle
Jul 24
Fanfic Writers Turn On Each Other as AI Detection Crusade Sparks Friendly Fire
Jul 5
Roku Launches Around-the-Clock AI Content Channel Amid Growing FAST Debate
17h ago