AI Product Building Future of AI Business Models

LLMs selectively destroy vertical software moats — 5 fall, 5 hold

Learned interfaces, custom workflows, public data access, talent scarcity, and bundling collapse under LLMs, while proprietary data, regulatory lock-in, network effects, transaction embedding, and system-of-record status remain defensible

@nicbstme — 10 Years Building Vertical Software: My Perspective on the Selloff · Mar 5, 2026 · 16 connections

Not all moats are created equal in the LLM era. Bustamante identifies five vertical software defenses that LLMs dismantle: learned interfaces (“LLMs collapse all proprietary interfaces into one Chat”), custom workflows where “years of engineering versus one week of writing” captures the shift to Markdown skill files may replace expensive fine-tuning, public data access that frontier models trivialize, talent scarcity inverted when domain experts encode methodology directly, and bundling broken by agents that orchestrate across providers.

Five moats hold firm: proprietary data that’s irreplaceable, regulatory lock-in, network effects, transaction embedding, and system-of-record status. When moats collapse, the strategic response shifts from selling tools to selling outcomes — Sell the work, not the tool — model improvements compound for services, against software argues that model improvements compound for services providers while eroding tool-sellers. This framework refines The UI moat collapses — API quality becomes the purchasing criterion — the UI moat is just one of five collapsing defenses. The resilient moats share a pattern: they’re rooted in data gravity or institutional trust, not in software complexity. This connects to SaaS survives as the governance and coordination layer — determinism still rules — the SaaS layer that survives is precisely the one anchored in these durable moats. One layer further up the stack is Organizational shape is the emerging moat in AI — what AI cannot copy is the institution underneath — when the product/data moats either fall or hold, the question of what kind of institution surrounds either case becomes the next-tier moat AI cannot copy.

Connected Insights

References (5)

→ Organizational shape is the emerging moat in AI — what AI cannot copy is the institution underneath → SaaS survives as the governance and coordination layer — determinism still rules → Sell the work, not the tool — model improvements compound for services, against software → Markdown skill files may replace expensive fine-tuning → The UI moat collapses — API quality becomes the purchasing criterion

Referenced by (11)

← In AI the threat is layer migration, not a competitor — work relocates across layers when any variable moves ← A frontier without an ecosystem is not stable — if a few models capture all value, the political economy won't tolerate it ← The system of work is the moat, not the model — the model is fungible underneath ← If a problem improves directly with raw model capability, the labs will take it ← Commodity work's terminal value is zero but structured expert judgment compounds indefinitely ← LLMs complete Aggregation Theory by collapsing the interface layer ← LLM competition fragments markets from 3 incumbents to 300 ← Sell the work, not the tool — model improvements compound for services, against software ← The comfortable middle is over — software companies must either accelerate AI growth or rebuild for 40%+ margins ← Self-disruption follows the value chain downward — software companies must eat their own agent layer before someone else does ← Organizational shape is the emerging moat in AI — what AI cannot copy is the institution underneath