Fireside Chat, Interview
Why a $1M Startup Wasn’t Worth Building in the Age of AI
- AI automation for small to medium-sized business tax preparation is expected to become feasible within the next two years, though incumbents are predicted to automate services for existing customers faster than new entrants can capture market share.
- AI adoption in personal contexts is anticipated to follow an iPhone-like curve within six to 18 months, while enterprise adoption of AI agents is projected to proceed at the speed of human adaptation, progressing from basic calendar integration to deep CRM access.
- Product-led growth will be the primary strategy for small and mid-market sectors, while security and compliance features will be developed in parallel for enterprise clients, with a long-term goal of offering 80% customized software per user.
- Market expansion is forecasted as token costs decrease, enabling the "long tail" of research tasks such as automated meeting prep notes, while a single-month monitoring agent for wholesale prices could generate hundreds or thousands in savings for small business owners against a monthly cost of approximately one dollar.
- The software value unit is predicted to shift from the application layer to "anything a human can do with a computer," causing app value to diminish as models improve, with expectations that agents will successfully perform tasks 90% of the time that were previously near-impossible.
- Neutrality in model selection is viewed as a critical requirement within three to five years, as it will become impossible to predict which provider will be superior, enabling customers to switch providers easily unlike the locked-in environments of competitors.
- Business efficiency is expected to improve significantly more than in the current environment within 12 months, transitioning toward a future with significantly less human intervention to realize economic benefits.
- New revenue models may emerge from the "breaking of covenants" regarding ad traffic, where businesses welcome agent traffic rather than blocking it, and the credit card benefits "breakage model" is predicted to die as agents optimize rewards for all users.
- Supply aggregation is expected to lose value as human attention is automated away, posing a risk to platforms that fight this shift, while businesses must find a "Goldilocks zone" of relevance to avoid being targeted by AI labs themselves.
- The future product experience is projected to move away from text boxes to proactive systems that perform actions like creating assets or managing workflows without explicit requests, reflecting the view that the AI market is currently underhyped outside of specific pockets.