Initialized Capital, Notable Capital, and V0 on Investing in Vertical AI

At the AI Agent Conference in New York, a panel of investors made one of the most direct arguments of the two-day event. Brett Gibson of Initialized Capital, Hans Tung of Notable Capital, and Christina Tan of V0 weren't there to discuss AI broadly. They were there to make a case: the next decade of AI value creation lives in vertical markets, not horizontal tools, and most organizations are still underestimating how deep that transformation goes.

Their thesis, stated plainly: AI is replacing workflows, not just adding features to them.

The Opportunity Is Workflow Replacement, Not Copilots

The panel was emphatic on this and returned to it throughout the session. The opportunity in vertical AI is not building a smarter assistant that sits alongside existing workflows. It's rebuilding the workflows themselves.

"The actual value is replacing workflows. AI lets you absorb operational labor directly."

Traditional SaaS organized information. It gave people better tools for doing work. Vertical AI performs the work itself. That's a categorically different value proposition — and a categorically different business model. "You have to do a lot more than just build software."

The investors framing this clearly: copilots and productivity plugins are above the line of what gets commoditized. Workflow replacement is where defensible businesses get built.

Legacy Software Is Worse Than Most People Admit

One of the most quotable moments of the session. The panel didn't hedge on this. "A lot of existing software is horrible. There are workflows that never had good software. These industries are still extremely inefficient."

This is the opening that AI-native startups are walking through. Incumbent enterprise software in construction, insurance, legal, compliance, healthcare, and accounting was built around human operators, manual processes, and UI paradigms that predate modern AI by decades. "The workflows were never designed for AI. These systems are deeply fragmented."

AI-native companies can bypass that entirely — redesigning workflows from first principles, eliminating operational handoffs, compressing labor layers that legacy systems assumed were permanent. The incumbents can't move fast enough to close that gap.

Start Narrow, Then Expand

The panel's practical advice for vertical AI founders was consistent and specific. The companies that win don't try to transform an entire industry from day one. They find the most painful workflow in the industry, solve it completely, and use that foothold to expand.

"Find one painful workflow. Start with a wedge product. Then expand into the full platform."

The wedge strategy works because it generates trust, accumulates workflow-specific knowledge, and creates proprietary operational data — all of which compound into defensibility over time. The industries they kept naming: construction, insurance, compliance, healthcare, legal, accounting, financial operations. Every one of them has workflows that are painful, document-heavy, and underserved by existing software.

Software and Services Are Converging

This was one of the more forward-looking arguments of the session and worth sitting with. The investors predict that many successful vertical AI companies will look like hybrids — part software company, part consulting firm, part BPO provider.

"The line between software and services is blurring. The software itself becomes operational labor."

When AI systems execute work rather than support it, the traditional distinction between software and services starts to collapse. A platform that reads contracts, flags risks, and generates compliance reports isn't just software — it's doing work that used to require a team. Pricing models, organizational structures, and go-to-market strategies all have to adapt to that reality.

The Moat Is Workflow Knowledge, Not the Model

The most important investment insight from the session, and the one with the most direct implications for how organizations should think about their own AI programs.

"The value comes from understanding the workflow. The domain context matters enormously."

Foundation models are not a moat. Every competitor has access to them. The defensible position is built from proprietary workflow knowledge, industry-specific operational data, and embedded process understanding that compounds over time as the system learns from real usage. The companies accumulating that knowledge now — through deep vertical focus and genuine workflow integration — are building advantages that foundation model improvements won't erase.

Regulated Industries Are the Biggest Prize

The panel repeatedly highlighted financial services and healthcare as the largest long-term vertical AI opportunities precisely because they're the hardest to enter. Compliance requirements, trust barriers, and operational conservatism slow adoption — but they also filter out shallow competitors.

"Compliance and accuracy matter enormously." The industries that demand the most from AI deployments are the ones that will generate the most value from them. Slower adoption doesn't mean smaller opportunity. It often means the opposite.

"We're not even scratching the surface. There are opportunities everywhere."

This session was presented at the AI Agent Conference 2026 in New York. Speakers represented Initialized Capital, Notable Capital, and V0.