Town vs Instinct vs GrokBot | Why the AI Assistant Market Is Not a Bubble

Strategic Positioning and Market Landscape

• Town identifies AI assistance as the current hottest category in Silicon Valley, with a specific focus on mainstream email and calendar automation rather than just "tinkerer" tools.
• Founder Jean-Denis (JD) confirms that his product is a top-three strategic priority for both Google and Apple over the next 12 months, creating direct competitive pressure.
• Town is positioned against competitors like Instinct and Grokbot, but JD asserts Town and Instinct target different strategies: Town aims for mainstream enterprise network effects, while Instinct appears to focus on consumer acquisition with potential subsidy models.
• JD predicts that in five years, users will trust AI agents to autonomously decide what personal and professional data to share with others without human intervention, a shift from current "human-filtered" silos.
• The primary "moat" identified by Town is a "network effect at the agent level," where agents from different coworkers can query each other to solve problems, making switching costs prohibitively high once a team is onboarded.

Product Pivot and User Metrics

• Town pivoted from a failed AI tax prep business to its current email/calendar assistant after finding immediate product-market fit via a quick prototype, bypassing extensive customer interviews.
• The company has been in the market for three months with a focus on mainstream users (ICP) rather than power users.
• Town reports a conversion rate of over 15% for free-to-paid users, which JD describes as "extremely high" for product-led growth (PLG) models.
• Pricing tiers are set at $15, $49, $99, and $199 per month, with the $15 plan having the worst unit economics (most subsidized) to drive adoption, while the $99 plan is the most profitable per user.
• The company targets a "100 million paying users" scenario to achieve a $100 billion valuation, believing that business-side AI usage will generate significantly higher revenue per user than consumer-side usage.

Economics, Infrastructure, and Model Strategy

• Town utilizes a multi-model routing strategy, selecting specific models for specific tasks (e.g., Gemini/OpenAI for images, 11 Labs for voice) rather than locking into a single provider to optimize cost and performance.
• Currently, the vast majority of workload runs on "frontier" models due to capability requirements, though JD anticipates a shift toward open-weight models for routine tasks like email labeling within 18–24 months.
• JD argues that companies cannot subsidize AI forever; charging users is essential to validate ROI and prevent "token maxing" where users burn compute on low-value tasks.
• The company proactively alerts users via email when they appear to be executing "rogue routines" that consume excessive tokens, aiming to build trust rather than hide usage costs.
• JD estimates that roughly 4–5% of an engineer's salary is currently spent on AI tooling (Devon, Cursor, Codex, Claude), a figure expected to rise as token costs decrease and agent efficiency increases.

Competitive Threats and Future Interaction

• The biggest fear for Town is cannibalization by legacy incumbents (Google, Apple, Meta) who own the device or distribution layer (e.g., WhatsApp) and can integrate agents deeply.
• JD views Apple as having a structural disadvantage due to its lack of cloud infrastructure and heavy reliance on on-device processing, which limits the agent's access to the global data needed for true agentic work.
• A significant disagreement exists internally regarding whether to pursue the "parent/family" vertical, which has strong product-market fit but lacks the high-willingness-to-pay and enterprise virality of the B2B core.
• JD predicts the future interaction will involve a single entry point for the user (one agent on the phone/computer) that manages different data silos (work vs. personal) for privacy, rather than multiple specialized apps.
• The company is currently building "Townies" (naming and branding the agents) to foster a human-like relationship and emotional connection, a strategic bet on emotional defensibility similar to Snapchat's structural moat.

Operational Insights and Hiring

• JD notes that the speed of AI development has outpaced human learning cycles, allowing competitors to copy features within weeks, which compresses the "learning advantage" startups traditionally enjoyed.
• Town's hiring process has evolved to skip formal interviews for candidates vouched for by trusted existing team members who have worked closely with them, relying on high-trust referrals for top talent.
• The company refuses to announce fundraising rounds alongside product PR, treating them as separate events to maintain credibility and avoid skepticism.
• JD rejects the idea of a pure ad-supported AI model, citing prohibitive token costs and the conflict of interest where AI recommendations could be skewed by advertiser incentives.
• For board composition, JD specifically seeks a member with late-stage CFO operational experience to manage the complexities of large-scale compute procurement and token economics.