Thanks, @drew1618t, for bringing us back to the central point:
A product is not a business.
AI can build some kinds of products – a single function product like you describe sounds like a great application for AI. @Smorgasbord1 earlier in this thread brought up a couple reporters building a dashboard similar to Monday.
Those are only products. There’s much more to building a business, let alone one that takes share from established players.
As someone who has built numerous software products over the years, and built large businesses based on those products, as well as invested in software businesses for many years, here’s what I think you can look for to help you distinguish between the software companies that will likely benefit from AI, versus the ones that are threatened by AI.
The analysis lens I suggest using comes from Hamilton Helmers’ Seven Powers, which talks about seven different ways that businesses build persistent competitive advantage. Here they are, along with my commentary for software businesses and AI:
-
Scale Economies: unit costs fall as volume increases. For SAAS businesses, the incremental cost of serving another customer is often thought to be $0 – but when you consider the payback period for customer acquisition costs for a typical SAAS business, it is far from $0. Payback of 2-4 years is not uncommon. This should serve as an indication that there are many more factors going into building a successful software business than just the cost of the software. That point aside, I don’t think this is a primary field of competition for existingplayer.SAAS vs upstart.ai. If anything, it would argue in favor of existingplayer.SAAS, with thousands or tens of thousands of existing accounts, where the CAC has been paid off already.
-
Network Economies: the product gets more valuable as more users join. This is an important part of my thesis with INTU: they have tens of thousands of small and midsize business customers, which means their AI engine can learn things that carry across businesses; and they can put that learning to work on behalf of all their customers. A new entrant will struggle to acquire the same learning (they have many fewer customers). That’s not the typical application of network economies but I think it’s relevant for AI learning. Note that it’s related to Process Power (below).
-
Counter Positioning: A newcomer adopts a business model that an incumbent can’t copy without cannibalizing their existing business. This is one of the central points that @Smorgasbord1 is making: if the cost of developing the software for a new entrant is very low, they can charge much less for their solution. Incumbents will struggle to match on price because they can’t jeopardize their existing revenue. This is certainly a risk – but there are often other considerations at work for customers besides just price.
-
Switching Costs: Customers incur real or perceived penalties—money, time, data loss, retraining, risk—when changing providers. An example: NOW empowers users throughout a customer organization to use it to build workflows. A competitor would have to make it worthwhile for the customer to retrain all those users and rebuild all the workflows that are automated via NOW. That’s a non-trivial task.
-
Branding: A trusted reputation built through consistent experience and messaging creates demand preference and often a price premium beyond functional features. Enterprise software businesses typically don’t build strong brands, but of the software businesses I’ve mentioned, INTU and DUOL both stand out for brand.
-
Cornered Resource: Exclusive access to a valuable asset—IP, talent, data, supply, distribution, licenses—that others can’t get on comparable terms. This is one of the primary attributes of the kinds of niche software businesses that the serial acquirers I mentioned earlier, like CSU, go after. My earlier example was about bus dispatch software for municipal transit operators. There are only so many people who know how these systems work within transit companies, or who can explain how the operator in, say, Calgary, wants to do things differently than the operators in Toronto or Ottawa. Detailed insider knowledge of a small industry is a perfect example of a cornered resource.
-
Process Power: Unique, often tacit ways of working that improve quality or lower cost and get better with scale and learning. Embedded in culture and routines, these processes are complex, path-dependent, and hard for competitors to copy. AI brings a lot to the Process Power arena – but it favors the incumbent, with access to all the customer’s historical data and the ability to bring insights to bear that a competitor can’t match. In a product shootout, bet on the contestant who brings the bigger gun. DUOL is a good example of this: they’ve been learning how to get users to persist, to keep coming back day after day, to use their product. Contrast that with the online training course completion rate of around 1%, which is typical for online courseware.
You have to know the businesses you’re investing in to apply this kind of analysis, but I know that many of the posters on this board are deeply immersed in the businesses they invest in. I find it’s generally quite helpful to think about my portfolio in these terms. If I don’t know enough about the business to do this kind of analysis, and I don’t have the strong curiosity to learn, that’s a good indication that it shouldn’t be in my portfolio.
At any rate, I hope it’s clear that while AI will have an impact on all software businesses, what kind of impact that will be is very much dependent on the individual business.