The Verge
AI Startups Are Acquiring for Personality, Not Just Power
AI commoditizes → personality and design become moats
Level 1
What Happened
Two notable AI acquisitions landed within days of each other. Midjourney, best known for its image-generation models, acquired Co-Star, the personalized astrology app that blends NASA data with AI to deliver daily horoscopes. Separately, Cognition, the startup behind the AI coding agent Devin, acquired Poke, a conversational AI assistant known for texting users like a friend, in a deal valued in the low nine figures.
Key Points
- Midjourney acquired Co-Star; Co-Star founder Banu Guler becomes Midjourney's chief design officer.
- Cognition acquired Poke for a low-nine-figure sum to bring its personality-driven interaction model to the Devin coding agent.
- Both deals signal acqui-hires of design and interaction talent, not just technology.
Sources
TechCrunch
Bloomberg
Level 2
Why It Matters
These acquisitions are not isolated deals. They reflect a structural shift in how AI companies think about competitive advantage as foundation models converge in capability.
Key Points
- As raw AI model performance converges, user experience and emotional resonance are emerging as the primary differentiators that drive retention.
- Midjourney's purchase of Co-Star is effectively an acqui-hire of a design-first team with a loyal, habitual-use audience, accelerating its push into consumer apps beyond Discord.
- Cognition's rationale for buying Poke is explicit: personality in AI tools increases user satisfaction and daily engagement, even in professional productivity contexts.
- Poke had over 100 million messages exchanged in three months but struggled to be profitable, illustrating that viral AI products often need larger infrastructure owners to survive.
- Both deals compress the timeline for AI companies to ship polished, consumer-grade products by absorbing teams that have already cracked engagement at scale.
Sources
The Verge
TechCrunch
Bloomberg
Level 3
What Changes
These deals reshape competitive dynamics across consumer AI, developer tools, and the broader startup ecosystem. The acqui-hire of interaction designers and product teams signals a new phase of M&A logic in the AI industry, one driven by user psychology as much as technical capability.
Sources
The Verge
TechCrunch
Bloomberg
winners
- Midjourney gains a proven consumer app team and a first-party design leader in Banu Guler, dramatically accelerating its app roadmap.
- Cognition's Devin agent gains a personality layer that could improve developer retention and daily active use in a crowded coding-agent market.
- Poke users benefit from faster, more reliable infrastructure backed by Cognition's models, resolving the latency and cost issues that plagued the standalone product.
- Co-Star's user base gains access to Midjourney's AI infrastructure, potentially enabling richer, more visually immersive astrological experiences.
losers
- Standalone conversational AI startups now face a harder path to independence, as infrastructure costs make profitability elusive without a larger acquirer.
- Competitors to Devin, such as GitHub Copilot and Cursor, face a differentiated rival that now combines technical depth with a more humanized interaction model.
- Discord, which currently hosts Midjourney's primary user interface, faces reduced strategic leverage as Midjourney accelerates its own native app development.
implications
- The AI M&A playbook is evolving from acquiring data or model IP to acquiring interaction design talent and loyal user communities.
- Consumer AI apps with strong engagement but weak unit economics are now viable acquisition targets for better-capitalized AI infrastructure players.
- Apple's Messages for Business platform, where Poke was the first approved AI agent, becomes a more contested and strategically important distribution channel.
- The line between a productivity tool and a social companion in AI is deliberately blurring, with implications for enterprise adoption norms and acceptable-use policies.
minority report
- Acqui-hires of culture-driven teams have a poor historical integration record: the personality and voice that made Co-Star and Poke distinctive were products of their independent, founder-led cultures, and absorbing them into larger organizations frequently dilutes exactly the quality that justified the acquisition price.
- Midjourney's move into astrology may reflect a lack of conviction in its core image-generation business rather than confident diversification, potentially signaling that growth in that vertical is plateauing faster than public narratives suggest.
Level 4
What Happens Next
The immediate operational picture is stable: both Poke and Co-Star continue as products through year-end. But the medium-term trajectory points toward deeper product fusion, new distribution bets, and intensifying competition for interaction design talent across the AI sector.
Sources
The Verge
TechCrunch
Bloomberg
second order
- Expect a wave of similar acqui-hires as AI infrastructure companies without strong consumer front-ends begin competing for the small pool of startups that have cracked habitual daily engagement.
- Apple's Messages for Business platform is likely to attract more AI agent applications following Poke's pioneering approval, turning it into a meaningful new distribution battleground alongside app stores.
- The fusion of Poke's orchestration model with Devin's coding capability could produce the first AI agent that manages multiple parallel software engineering sessions under natural-language instruction, a meaningful leap beyond current single-task agents.
prediction
- Midjourney will ship its first standalone consumer app within 12 months, with Banu Guler's design influence making it visually and experientially distinct from existing text-to-image interfaces.
- Poke and Devin will begin partial product convergence in 2026, with Poke acting as a persistent memory and orchestration layer across Devin coding sessions, effectively creating a co-worker that never forgets context.
- At least two more top-10 AI labs will announce similar personality or design acqui-hires within the next two quarters as the interaction-quality arms race accelerates.
minority report
- Regulation targeting AI companions and emotionally engaging agents, already advancing in the EU under the AI Act's provisions on psychological manipulation, could materially constrain the personality-driven AI model both acquisitions are betting on, particularly for applications targeting younger or vulnerable users.
- Cognition's angel investment history in Poke, disclosed by co-founder Scott Wu, raises governance questions about deal pricing and fiduciary obligations to outside investors that could attract scrutiny and complicate future fundraising narratives.
Level 5
What This Means
For operators, investors, and founders, these two deals crystallize a strategic thesis that has been building quietly: in a world where model capability is increasingly commoditized and API-accessible, the durable moat shifts to the layer that sits between the model and the human. That layer, which encompasses tone, memory, personality, timing, and ritual, is now being priced and acquired like hard infrastructure.
What This Means
Personality is now a product category, not a feature.
AI Product Strategy
Both acquisitions confirm that interaction design, emotional cadence, and habitual engagement loops are distinct and acquirable capabilities, not emergent properties of better models. Founders building in this space should articulate and protect their interaction model as core IP, not just their training data or architecture.
High-engagement, low-margin AI apps are viable acquisition targets even without a path to standalone profitability.
Venture Capital
Poke's 100 million messages and hundreds of thousands of users could not generate sustainable unit economics alone, but they were precisely what made it valuable to Cognition. Investors should re-examine portfolio companies through the lens of strategic acquirability, not just standalone IRR.
The humanization of developer tooling is no longer a consumer conceit.
Enterprise and Developer Tools
Cognition's explicit rationale, that developers prefer co-workers with personality over robotic tools, challenges the long-standing enterprise assumption that professional software should be neutral and frictionless. Expect human-centered design language to enter procurement conversations for coding agents, DevOps tools, and knowledge management platforms.
Detected Trends
Personality as Moat
interaction-design-as-IP
AI companies are acquiring for tone, memory, and emotional engagement, treating interaction models as defensible infrastructure rather than surface-level product polish.
Acqui-hire 2.0
ai-acqui-hire
The M&A logic has shifted from acquiring model weights or data to absorbing engaged user communities and the small teams who built habitual-use behaviors into their products.
Consumer AI Consolidation
consumer-ai-consolidation
High-engagement, low-margin AI consumer apps are being absorbed by infrastructure-rich acquirers, compressing the window for standalone consumer AI companies to reach profitability independently.
Sources
The Verge
TechCrunch
Bloomberg
implications
- AI companies that have not yet built a consumer engagement surface face a narrowing window to acquire one before valuations for proven interaction-first teams escalate further.
- The Co-Star acquisition positions Midjourney in a daily-use ritual category, something image generation has never achieved, which could fundamentally change its revenue model from episodic to subscription-habitual.
- Founders with strong interaction models and engaged communities should treat inbound acquisition interest as a pricing signal for their next fundraise, not just an exit option.
second order
- If Midjourney's app layer succeeds, it becomes a vertically integrated AI consumer company with models, design, and distribution in-house, a structure no current AI-native company has fully achieved, intensifying pressure on OpenAI and Anthropic to follow suit.
- The normalization of AI agents with persistent memory and personality in both consumer and professional contexts will accelerate regulatory scrutiny of data retention, emotional dependency, and disclosure requirements, creating compliance overhead that further advantages well-resourced incumbents over startups.
minority report
- There is a credible case that the interaction-quality thesis is overstated: enterprise buyers have historically deprioritized personality in tooling in favor of security, compliance, and integration depth, and the developers most likely to use Devin are precisely the users least persuaded by conversational warmth as a value driver.
- Midjourney's move into astrology, a domain with no defensible data moat and intense competition from free apps, may ultimately prove a distraction that delays its core image-generation app while better-funded rivals like Adobe Firefly and Google Imagen ship native mobile products first.