Workflow Grit as Vertical AI Moat — The Messier the Market, the Deeper the Defensibility

The two hiding forces

Workflow grit. Exceptions, legacy system integrations, human approval loops, compliance requirements, expensive failure modes. When a company builds AI into a gritty workflow, it produces far more than model access: structured messy data, integrated old systems, designed approval loops, defined acceptable error rates, won trust where mistakes are costly. That work compounds into a proprietary map of how the workflow actually runs. No new entrant replicates it by buying the same model.

Market structure. Fragmented across thousands of small operators with low tech DNA. Horizontal AI vendors need concentrated high-value customers. Vertical systems willing to go market by market make those economics work. The buyers — property managers, CPA firms, outpatient clinics, field services companies — lack the engineering instinct to build production AI. They buy systems that arrive ready to do the work.

Together: the grit hides the market and creates switching costs. The fragmentation keeps horizontal players out. The low build instinct widens the window. The vertical system compounds before anyone else shows up.

The Goldilocks TAM

Standard TAM sizing uses software spend as the denominator. In operationally messy verticals, software budgets are modest. But the real money sits in services and labor.

The expansion happens when the product shifts from assisting work to doing it. EliseAI started at ~$50K ACV for leasing automation. Today: millions per customer across leasing, maintenance, collections, and AI-guided tours. The TAM didn’t grow. The product revealed how large it always was.

The Goldilocks zone: large enough for a venture outcome but disguised by modest software surface area. Too-obvious markets attract labs directly. Hidden markets give the vertical system years to compound before anyone notices.

Models → Wedges → Systems

Sapphire Ventures frames the progression:

“Models win demos. Wedges win pilots. Systems win markets.”

Two trajectories splitting vertical AI companies:

• Vertical AI product (intelligence injected into a task) — demos well, deploys fast, gets commoditized when the model improves or a competitor copies the feature
• Vertical AI system (data, integrations, approvals, exceptions, compliance, labor budget owned end-to-end) — slower to build, harder to remove, compounds into the way the customer operates

Connection to existing thesis

This reinforces and extends several existing notes:

• Run the Business vs Do the Work - The AI-Era Vertical SaaS Shift — “do the work” products are the wedge; the system position is what you build after the wedge lands
• Hero User Strategy - Native AI’s Wedge Into Vertical Software — the hero user is the entry point into the gritty workflow
• Neofirms - AI-Native Professional Services as a New Category — neofirms are what the system position looks like at maturity (agents do the work, humans own quality)
• Full Stack Down vs Full Stack Up - Two Directions for AI Application Vertical Integration — Full Stack Up into service delivery is the path to capturing the labor budget
• The Agent Is the Customer - A Convergence Thesis on Where AI Value Accrues — the system position is what wins in the Category B incumbent markets
• Distribution as the Remaining Moat - Why SaaS Incumbents Aren’t Dead — workflow grit creates a different kind of moat than distribution. Category A incumbents hold on distribution; vertical AI systems hold on operational context.

Related Notes

• AI + Business Model — parent hub
• AI and Investing Thesis — investment implications
• Healthcare Admin Automation - Data Transformation and Vertical Workflows — healthcare as a gritty-workflow example
• Sapphire Ventures — The Biggest Vertical AI Markets Are Hiding in Plain Sight — source
• The Relational Sector - Why Human Involvement Becomes the Product — relational sector work is inherently gritty; human involvement is the moat