Fireside Chat, Interview
Why a $1M Startup Wasn’t Worth Building in the Age of AI
- Pivot Decision: Jean Denis and his co-founder pivoted Town from a SMB tax automation product to a general AI assistant after one year, recognizing that tax is an annual, low-frequency task with high barriers to hyper-scaling and vulnerable to incumbent competitors.
- Market Constraints: The previous tax venture faced "discontinuity" issues: interactions occur only once a year, preventing daily active user (DAU) metrics, viral word-of-mouth, and rapid market penetration before incumbents could replicate the technology.
- New Direction: The new Town product focuses on autonomous AI agents for common knowledge worker workflows, specifically email management, meeting preparation, and calendar organization, leveraging model improvements (e.g., Opus 4.5) to enable deeper agency.
- Go-to-Market Strategy: Town is adopting a Product-Led Growth (PLG) model targeting SMBs and mid-market firms first, prioritizing user experience over enterprise compliance features to mimic consumer-grade adoption before addressing complex enterprise security later.
- Valuation Logic: The product operates on a "token allocation machine" model, where AI identifies low-value, high-volume tasks (e.g., draft meeting notes, analyze wholesale price changes) that cost pennies to automate but save users significant time or hundreds/thousands of dollars annually.
- Competitive Positioning: Town differentiates itself from major AI labs (OpenAI, Anthropic) through three core tenets: neutrality (using the best model provider at any given moment), data ownership (users retain full rights to their context and memories), and a refusal to train on customer data for third-party resale.
- Strategic Vision: The company aims to replace the "app" layer with a unified, customizable AI agent that learns user-specific workflows, arguing that the future value unit is "anything a human can do" rather than discrete software features.
- Forward-Looking Market View: Jean Denis predicts that the market will eventually shift from adversarial blockages of agent traffic (e.g., Amazon blocking scrapers) to synergistic integration, where supply aggregators must adapt to agent-mediated discovery or face obsolescence.
- Adoption Barrier: The primary bottleneck for AI diffusion is not intelligence availability but product design; the market requires proactive, non-text-box experiences that "show value" rather than requiring users to be "tinkerers" to solve their own problems.
- Internal CRM Concept: Town is developing a privacy-preserving internal CRM mode where agents can query across employee data to find connections (e.g., "Who knows someone in the finance team?") without human-in-the-loop bottlenecks.
- Future of Work: The future of corporate efficiency will involve significantly less human-in-the-loop interaction, with AI mediating security and privacy to allow immediate, automated responses to internal queries that currently rely on slow, human-mediated channels like Slack.