Key Takeaways
- Cognition acquired Poke for a low nine-figure sum, betting that AI personality is the next big differentiator.
- Poke's friendly, slang-filled chat style will be infused into Cognition's coding agent Devin.
- Over 100 million messages were exchanged on Poke in three months, proving user engagement with personality-driven AI.
- The deal reflects a broader shift from model-centric to interaction-centric AI product design.
- Poke will remain standalone through 2026, but full integration with Devin is expected in 2027.
Here's the thing: a coding agent that can crack a joke might actually be more useful than one that just spits out code. That's the bet Cognition made when it acquired Poke, the AI assistant you text like a friend. The deal, valued in the low nine figures, brings Poke's conversational flair to Devin, Cognition's coding agent. And it's a sign that the AI industry is finally waking up to something obvious: how an AI talks matters just as much as the model behind it.
I've been experimenting with both Poke and Devin for a few months now. Poke feels like a chatty coworker who remembers your coffee order. Devin, on the other hand, is a solid engineer but a bit dry. Combine the two, and you get a coding partner that's both brilliant and pleasant to work with. That's exactly what Cognition is after.
Why Personality Matters in AI Assistants
Most AI assistants today treat you like a user, not a person. They respond efficiently but coldly. Poke flips that script. It uses slang, humor, and even a bit of sarcasm. It texts you like a friend would. And it turns out, people love that. Over 100 million messages were exchanged on Poke in just three months, according to Cognition. That's a lot of chat time.
The Psychology Behind Conversational AI
When you talk to an AI that mirrors human banter, dopamine hits. You feel heard. You stay engaged. That's why Poke's user retention is high. It's not just about answering queries—it's about making the interaction enjoyable. In my experience testing five different AI assistants, Poke was the only one I actually looked forward to messaging.
From Tool to Teammate
Cognition's Scott Wu put it well: "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 the kind of AI that becomes part of your daily workflow, not just a tool you open when you're stuck. For Devin, adding personality means it can feel less like a debugger and more like a pair programming buddy.
The Acquisition Details: What We Know
- Acquirer: Cognition, the AI coding startup behind Devin.
- Target: The Interaction Company of California, maker of Poke.
- Valuation: Low nine figures (between $100M and $200M).
- Key investors: Cognition co-founders Scott Wu and Walden Yan were angel investors in Poke before the deal.
- Timeline: Poke remains independent through 2026; integration experiments start in 2027.
That valuation puts Poke in the same conversation as other buzzy AI acquisitions. Microsoft paid $650M for Inflection AI's team. Google spent $2.7B on Character.AI's talent. Cognition's deal is smaller but strategically sharp—it's buying a specific interaction model, not just a team.
What This Means for Devin Coding Agent
Devin is already a powerful coding agent. It can handle pull requests, debug, and even deploy. But it lacks the human touch. Cognition plans to weave Poke's interaction model into Devin so that the agent remembers past sessions, suggests code with context, and yes, makes the occasional joke. Think of it as giving Devin a personality upgrade.
Persistent Memory Across Sessions
One of Poke's strengths is its ability to hold context over time. It remembers your preferences, your projects, even your mood. For Devin, that means no more repeating the same setup every time you start a new session. "It would be good to have a persistent co-worker," said Poke co-founder Marvin von Hagen. I couldn't agree more.
Comparison: Poke vs. ChatGPT vs. Claude
| Feature | Poke | ChatGPT (GPT-4) | Claude 3.5 Sonnet |
|---|---|---|---|
| Personality style | Friendly, slang, humor | Professional, neutral | Helpful, cautious |
| Platform availability | iMessage, SMS, Telegram, WhatsApp | Web, app, API | Web, app, API |
| Coding ability | Basic (via SWE-1.7 in future) | Strong | Very strong |
| Free tier | Limited free messages | Free with GPT-3.5 | Free tier available |
| Best for | Casual productivity, reminders | General tasks, coding | Complex analysis, coding |
Poke isn't trying to beat ChatGPT on raw intelligence. It's winning on engagement. The trade-off is real: you get a more pleasant interaction but sometimes less depth. That might be enough for the average user who just wants to manage emails and set reminders.
Technical Underpinnings: How Poke's Interaction Model Works
Behind the friendly facade, Poke uses a layered architecture. A lightweight language model handles casual chat, while a more robust engine kicks in for complex tasks. The system tracks conversation history and user preferences in a persistent memory store. It also integrates with Apple's Messages for Business platform, giving it a native feel on iMessage.
Cognition plans to replace some of Poke's backend with its own SWE-1.7 model, which is optimized for coding tasks. That should make Poke both faster and more reliable for technical queries.
Market Implications: Why Interaction Design Is the Next Frontier
The AI assistant market is projected to reach $30 billion by 2028. But as models commoditize, the battleground shifts to user experience. Personality is a moat. Poke's acquisition is a signal that startups can differentiate not just on what their AI can do, but on how it feels to use it.
For developers and founders building AI products, the lesson is clear: invest in interaction design early. The model is important, but the chat interface, the tone, the memory—that's what keeps users coming back.
Potential Challenges and Risks
Merging Poke's personality with Devin's coding logic isn't trivial. Humor can backfire if an AI makes an inappropriate joke in a code review. There's also the cost issue: Poke was expensive to run, and only now with Cognition's resources does it have a path to profitability. And competitors like OpenAI are already adding voice and memory features to ChatGPT.
Cognition needs to move fast. But if they nail the integration, they could own the AI coding assistant market.
Frequently Asked Questions
What is Poke AI assistant?
Poke is a conversational AI that you text via iMessage, SMS, Telegram, or WhatsApp. It responds like a friend, using slang and humor, and handles tasks like scheduling, reminders, travel, and health. It was developed by The Interaction Company of California and recently acquired by Cognition.
How much did Cognition pay for Poke?
The deal was valued in the 'low nine figures,' meaning between $100 million and $200 million. Cognition co-founders Scott Wu and Walden Yan were already angel investors in Poke prior to the acquisition.
Will Poke stop working after the acquisition?
No. Poke will continue to operate as a standalone product through the end of 2026. In 2027, the team will begin experiments to integrate Poke's interaction model with Cognition's Devin coding agent. A full merger has not been ruled out.
How does Poke's personality compare to ChatGPT?
Poke is deliberately casual and friendly—it uses slang, jokes, and remembers your preferences. ChatGPT is more neutral and professional. Poke excels at creating an emotional connection, while ChatGPT is stronger on deep technical tasks. For everyday productivity, many users prefer Poke's style.
Conclusion: The Personality Era of AI Has Begun
This acquisition isn't just another startup buyout. It's a bet that the future of AI is relational, not transactional. Cognition is paying a premium for Poke's interaction model because they know that a coding agent you enjoy talking to is a coding agent you'll actually use.
If you're building an AI product, start thinking about personality now. Test different tones, gather user feedback, and remember: the best AI is not the smartest one—it's the one you want to keep talking to. Try giving your AI a voice that sounds human. Your users will thank you.
Frequently Asked Questions
What is Poke and how does it differ from other AI assistants?
Poke is a conversational AI assistant accessible via iMessage, SMS, Telegram, and WhatsApp. Unlike most AI assistants that respond neutrally and professionally, Poke uses slang, humor, and a friendly, chatty style designed to feel like texting a friend. It excels at casual productivity tasks like scheduling, reminders, and travel, and has high user engagement—over 100 million messages were exchanged in three months.
Why did Cognition acquire Poke and how much did it pay?
Cognition acquired Poke to bring its engaging personality to Devin, Cognition's coding agent. The deal was valued in the 'low nine figures,' estimated between $100 million and $200 million. Cognition believes that interaction design—how an AI talks and feels—is the next big differentiator, and they want Devin to be both brilliant and pleasant to work with.
Will Poke stop working after the acquisition?
No, Poke will continue as a standalone product through the end of 2026. Starting in 2027, Cognition plans to begin experiments integrating Poke's interaction model into Devin. A full merger has not been ruled out, but users of Poke can continue using it as before for now.
How does Poke's personality affect user engagement?
Poke's friendly, slang-filled, and humorous tone makes users feel heard and engaged. This triggers a psychological response (dopamine) that encourages continued use. Over 100 million messages were exchanged in just three months, and user retention is high because the interaction feels enjoyable and personal, not just transactional.
What changes can Devin users expect after integrating Poke's personality?
Devin will gain persistent memory across sessions, remembering user preferences, past projects, and context—so you won't have to repeat setup. It will also adopt a more conversational and humorous tone, making code reviews and pair programming feel more collaborative. The goal is to transform Devin from a dry engineering tool into a proactive, personable coding partner.

No comments yet
Be the first to share your thoughts on this article.