AI is splitting software into the exposed and the entrenched. We advise companies built to capture the shift, not be captured by it.
The next few years will separate companies with structural defensibility from those whose advantage was simply being first to build the obvious thing. We look for the former, and we price a sale on the strength of the moat, not the size of the feature set.
The list below is what we underwrite. Most great companies have one or two. The rare ones compound several.
Six moats that survive AI. Each holds for a structural reason, and each shows up in the financials as pricing power and retention.
Scarce, hard-won data that compounds with every customer and can’t be scraped, bought, or synthesized.
As models commoditize generic knowledge, value migrates to the proprietary inputs they can’t reach. Whoever owns that data sets the limit on how good anyone’s product can get.
A decade of labeled outcomes, exclusive supply relationships, or a dataset competitors cannot reconstruct at any price.
The compliance burden that protects the incumbent: rules that make switching expensive and new entry slow.
AI lowers the cost of building software, not the cost of being audited. Regulated workflows stay sticky because the risk of getting them wrong is asymmetric.
Certifications, audit trails, and approval cycles embedded in the product; customers who can’t switch without re-certifying.
Value that grows with every participant: a network no new entrant can cold-start.
A better interface doesn’t move a two-sided market. The liquidity is the moat, not the UI, and liquidity is the hardest thing to copy.
Marketplaces, multi-party workflows, and data that improves for everyone as each new customer contributes to it.
Software that sits in the flow of funds: payments, billing, or capital running through the product.
Once money moves through you, you’re load-bearing. Rip-and-replace risk climbs, and monetization compounds as the customer grows.
Embedded payments, take-rate revenue, or working-capital and credit products underwritten on proprietary data.
The database the business runs on: where the canonical data lives and every other tool reads from.
AI may capture the interface, but the system of record owns the data underneath. The agent still has to write back to you.
The ERP, PMS, or ledger of an industry: the source of truth that downstream applications integrate against.
The operational layer customers build around: not an app they use, but the rail everything else runs on.
Control points accumulate workflow gravity. As an ecosystem forms around them, displacement means rebuilding everyone’s integrations, not just one vendor’s.
The platform third parties plug into; the workflow others extend rather than replace.
Moats are the outcome. Control points are the mechanism: the specific ways vertical software earns a position competitors can’t dislodge. We use these eight patterns to pressure-test how durable an advantage really is.
Software that brings the customer more revenue: new demand, better conversion, or additional sales channels.
Embedded credit, insurance, or working capital that gives customers financial access.
The vendor shares in customer success through payments, take rates, and revenue-linked products.
Improving the end-customer experience across a fragmented value chain.
Don’t rip out the legacy system. Integrate, extract its data, and build the modern layer around it.
A cross-stakeholder source of truth spanning suppliers, employees, consumers, and regulators.
AI moves the software from administering a workflow to actually executing the task.
Third parties and industry participants build directly on the vendor’s platform.
30 minutes with the senior team. We’ll tell you which of these you actually have, and what it’s worth to the right buyer.
Schedule a Working SessionThe deal terms that quietly move value, monthly, from Shane. No spam, no marketing, no padding. Just the writing that founders actually keep.