TechCrunch→ original

Cognition Buys Poke: Why an AI Assistant's Personality Became a Competitive Advantage

Cognition, the maker of the Devin coding agent, has bought Poke, a startup whose AI assistant users text like a friend. According to TechCrunch, the deal values Poke in the low nine figures (over $100 million). Cognition is taking Poke's conversational style and interaction model and bringing them to Devin. The logic of the purchase: HOW an assistant talks to the user is becoming at least as important as the model under the hood.

AI-processed from TechCrunch; edited by Hamidun News
Cognition Buys Poke: Why an AI Assistant's Personality Became a Competitive Advantage
Source: TechCrunch. Collage: Hamidun News.
◐ Listen to article

Cognition, the startup behind the Devin coding agent, announced on July 24, 2026 that it has acquired Poke, an AI assistant that users text like a friend in a messenger. According to TechCrunch, the deal values Poke in the low nine figures, meaning more than $100 million.

What exactly Cognition bought

Cognition acquired less a technology than a way of communicating. Poke's main asset is its conversational style and interaction model: the user chats with the assistant as with a real person, rather than typing commands into a prompt line.

  • Buyer: Cognition, an AI startup and the creator of the Devin coding agent
  • Subject of the deal: Poke, an AI assistant in a personal-messaging format
  • Valuation: low nine figures (over $100 million), according to TechCrunch
  • Announcement date: July 24, 2026
  • Goal: bring Poke's communication style to the Devin agent

Why Devin needs it

Cognition is moving Poke's conversational interface into Devin, its agent for writing code. Devin is known as an autonomous tool that takes a task and completes it on its own; now Cognition is adding a more human way of interacting with the developer.

The logic is simple: even the strongest coding agent is useless if talking to it feels like filling out a technical form. Poke gave Cognition a ready-made conversational "shell" that would have taken months of in-house development to refine.

Why an assistant's personality matters more than the model

The deal reflects a shift in the industry's priorities: HOW an assistant talks to a person is becoming at least as important as the model under the hood. When the base LLMs of all the players are comparable in strength, the competitive advantage comes down to communication style: tone, memory of context, the feeling of talking to a living interlocutor.

"How AI assistants interact with users is just as important as the models they run on," is how

TechCrunch sums up the trend in its analysis of the deal.

This explains why Cognition paid a nine-figure sum essentially for the "character" of a product, not for a unique model or a patent. Personality is turning into an asset that can be bought and built in.

What this means

The Poke acquisition is a signal that the AI assistant market is entering a phase where model size does not decide the outcome; the quality of the interaction does. In the coming months, expect other labs to start buying up teams that know how to make conversations with AI feel lively and natural.

Frequently asked questions

How much did Cognition pay for Poke?

The exact amount has not been disclosed. According to TechCrunch, the deal values Poke in the "low nine figures", meaning more than $100 million but closer to the lower end of that range.

What is Poke?

Poke is an AI assistant that the user talks to in a personal-messaging format, like a friend in a messenger. Its key feature is its conversational style and interaction model, not commands and prompts.

ZK
Hamidun News
AI news without noise. Daily editorial selection from 50+ sources. A product by Zhemal Khamidun, Head of AI at Alpina Digital.
✦ Autonomous AI Agents

Delegate Deep Research, Coding & Workflows to an AI Agent

Deploy autonomous multi-step execution with NomadNet Agent (agent.nomadnet.ai) — automated web research, data synthesis, code generation, and complex task orchestration.

Launch NomadNet Agent →NomadNet AI Platform
What do you think?
Loading comments…