439: The Increasing Risk of Building in Public

What It Covers

Arvid Kahl, who sold FeedbackPanda using radical transparency in 2018-2019, argues that AI-powered agentic coding tools have collapsed the safe threshold for building in public from $20-30K MRR to effectively zero.

Key Questions Answered

• •Copycat threshold collapse: The previously safe zone for sharing business details publicly—below $20-30K MRR—no longer exists. Agentic AI tools let non-developers clone a business in days using a single prompt that scrapes social history, analyzes landing pages, and generates production-ready code.
• •The safe sharing filter: Apply a two-part test before posting anything publicly: will it generate genuine audience engagement, and does it avoid revealing cloneable specifics? Share unexpected obstacles, general industry observations, and bug-fixing experiences—never revenue figures, customer counts, or system architecture.
• •Protect infrastructure details: System architecture, data pipeline configurations, and specific third-party service dependencies represent hard-won operational knowledge built over years. Sharing these publicly hands competitors an actionable blueprint that agentic tools can immediately translate into a functional competing product.
• •Relationship moat as defense: Customer relationships built through six-plus months of emails, team meetings, and negotiations remain one defense AI cannot yet replicate at scale. Early-stage founders should prioritize building these relationship moats while keeping product and technical details private.