Business Models vs. Verticals, News From The Front & Required Reading
Capital Efficient #2
I’m Pat McGovern and welcome to the second edition of Capital Efficient. We are still experimenting with formats here at CEHQ so expect the unexpected over the next few weeks as we find our groove.
Weekly Radar
Vertical AI → Business Model AI
Driving IPO-scale outcomes in the land of vertical software isn’t easy - this is evidenced by the fact there are only about ~20 public vertical SaaS co’s founded in the last 25 years. Post-GPT, interest in vertical software has exploded, and now founders are slicing their verticals thinner and thinner to stand out from dozens of venture-backed competitors (e.g., we aren’t software for truckers; we’re software for refrigerated box-truck operators on the West Coast.) While this can make you distinct, it can also cap your addressable market.
I’m increasingly interested in vertical software that tackles an entire business model from Day One rather than locking into a narrow sub-industry. While each industry has its nuances, certain types of vertical software are more generalizable to adjacent categories than founders often admit — and a larger initial customer base can be a big advantage. Our portfolio company Harmonyze is a great example of this - they are building for franchisors (which is more of a business model than a single vertical as this encompasses everything from QSRs to gyms to hardware stores and swim schools). Another example would be building an OS/command center for PE-backed businesses rather than just for car washes or for laundromats or for fast food restaurants, etc.
n.b. - this idea of business model software goes against the typical advice of start very narrow and expand outward; but for Vertical AI given the potential speed of product development (and expectations around speed of revenue by Series A/B investors), it may be worth considering a broader initial GTM target base. If you do start narrow, have a plan to widen your target fast - not in 2-3 years. Also, for really complex workflow automation (e.g., prior auth for providers/payors, this is far less applicable).
This phenomenon of “business model AI” can also be seen in the roadmaps of Hebbia, Harvey, and others who are evolving from “AI for financial firms” and “AI for law firms” to the more generalizable “AI for Professional Services” approach. Harvey’s public ICP is AmLaw 250 Law Firms but PwC and other consulting firms have become some of their most important customers. Hebbia has also branched out from finance into Big 4 consulting. This land grab by Vertical AI startups into industries with similar business models is one I expect to continue.
News Roundup
Figma’s IPO Delivers
As predicted last week, outsized demand for new tech IPOs coming to market led to a huge pop in Figma’s first day, with trading closing at $122/per share and as of today at noon, the stock was still at ~$80/per share (more than 2x above where banks had initially priced it). While it’s great to see the IPO window nominally open, companies still need to be doing several hundred million in revenue with a 40% YoY growth narrative to excite the public markets and these are relatively few and far between among today’s late-stage, venture-backed private co’s.
Harvey Crosses $100MM ARR
Harvey announced this week that it crossed $100MM ARR. This aligns with my observations that there are two kinds of AI-native software companies that have demonstrated an ability to scale very quickly
Startups that sell low-cost, PLG to thousands of prosumer users (e.g., Cursor) and;
Startups that sell very high-end platforms to the blue-chip players in their category (e.g., Harvey for Legal, Abridge for Health).
To date, I haven’t seen as many AI-native companies serving the mid-market that have exhibited this kind of explosive growth from $0 → $100MM ARR but that may change as AI creeps into more of the economy and becomes table stakes for all businesses.
The Next Brex vs. Ramp
A new rivalry is taking shape in the land of enterprise resource planning. Rillet - an AI-native ERP - announced this week it raised $100M from a16z, ICONIQ, Sequoia, and others. This comes on the heels of DOSS raising $18MM in a Series A led by Theory Ventures in April 2025. I predict DOSS and Rillet will follow an analogous trajectory with VCs pouring cash into each of them as they publicly duke it out to be the one, true AI-native ERP. In my view, the market for ERPs is a big one and multiple winners will emerge.
What I Am Reading
“SEO is Dead. Say Hello to GEO.” (NY Mag) - John Herrman with an overview of the challenges the SEO-centric web will face as AI tools replace search engines for information discovery. The future of ad-tech (and the open web) will be written by the decisions the foundation model labs make over the coming years.
“Bessemer Venture Partners (2025)” (HBR) - Earlier this summer, Harvard Business Review released a case study on the multi-stage firm’s expansion into Seed. Interestingly, the case shows Kent Bennett citing Andreessen’s push into Seed as a forcing function for BVP doing the same. Good read for any pure-play Seed fund trying to understand multi-stage motivations.