AI start-ups: Don’t sell picks and shovels, dig with them
AI is going to change everything. At least, that’s what everyone in Startup-land believes.
The rapid development of new foundation models and jaw-dropping demos are exciting. Unfortunately, the velocity of new technology development, the lack of clarity around legal issues, the general accessibility of the models, and the unique commercialization challenges posed by AI make founders unsure how to build startups in this new landscape.
Many founders think the winning move is to build picks and shovels for the latest tech gold rush. The logic is that the safe (and smart?) space was building tools to help others build out end-products in previous tech waves. This thesis is a perfectly defensible strategy that worked well during the dot-com boom, web 2.0, and into the mobile era.
I think this time is different, and the playbooks that have served startups so well over the last 15 years aren’t the ones that will help them win in 2024 and beyond.
SaaS = Selling tools instead of solutions
The last decade and a half of B2B software development has been marked by nonstop development of new tools. These were largely point solutions that helped marketers, CS managers, HR folks, and other employees gain insights, improve workflows, and streamline communications internally and externally.
Even the people making tools had new tools! Horizontal low-code apps like Airtable and Shopify turned small businesses and office workers into pseudo-engineers. Underlying those was another layer of “picks and axes” like Plaid and Zapier, making the interconnection between apps easy to build. Additionally, the rise of cloud providers commoditized hosting.
Don’t sell accounting firms more SaaS. Replace them!
Many founders are trying to copy the SaaS playbook in the AI era. They should consider reframing their thinking around full-stack businesses or a new generation of tech-enabled services. Instead of selling companies on a new AI-powered accounting software package for their workers, imagine a firm built from the ground up with AI tools at heart for digesting financial statements, auditing questionable entries, and producing detailed financial reports.
Or imagine a new law firm, real estate brokerage, consulting firm, or healthcare practice doing the same. Perhaps Atrium — which was trying to reinvent the law firm several years back — was simply too early.
I spoke with a founder who built and sold a healthcare business requiring a large call center to talk to customers and process insurance claims. He told me that most of those facilities today would be replaced by AI, and the business margins would have been much higher. This example is just one of many potential optimizations founders could capitalize upon.
If you have a massively successful and proprietary algorithm for trading stocks or bonds, would you sell the tool or trade stocks and bonds?! Clearly, you would do the latter.
But won’t existing firms adopt AI?
Theoretically, legacy businesses should be able to leverage AI tools. The benefits are apparent and generally easy to use and adopt.
However, it takes a lot of work for legacy organizations to retool at a fundamental level. SaaS tools helped people modernize historical processes — AI may eliminate the need for most of them. Naturally, many employees will become defensive of their responsibilities and protective of their jobs as AI encroaches. Adoption may be slower than expected. This reality creates an opportunity to set up a business with AI at the core and compete with laggards.
Why you should build a full-stack startup powered by AI
Hesitation to fully adopt and understand AI’s potential at legacy companies is an opportunity for startups to thrive. It’s not easy to build a new Deloitte or McKinsey. Much of what these firms sell besides the work product is a level of trust.
Still, many clients of small and mid-market service firms don’t need or can afford this brand equity, which can be a big enough wedge for a startup. Servicing these clients can create an opportunity for a new generation of service businesses that, over time, may achieve the level of trust and brand recognition of the incumbents, perhaps even more so.
Don’t VCs hate tech-enabled services?
Historically, a full-stack, service-oriented approach to company building has been unpopular among founders. Entrepreneurs believe, accurately, that VCs don’t like to invest in tech-enabled services businesses. Full-stack startups are also hard to execute, margins are lower, and they require more people instead of pure software.
That said, if you believe AI is the game-changing technology it’s said to be, these objections may no longer hold. You may only need a fraction of the workforce you did previously, so margins should improve dramatically. Full-stack solutions are a risky bet, but no riskier than trying to sell the 501st shovel in a crowded market. Being yet another tool isn’t a great moat in this new world. Defensibility will arise from building a new end-to-end solution that does the jobs by leveraging AI from the ground up.
Of course, there will always be a market for new software companies. All I suggest is that founders consider some out-of-the-browser alternatives. No matter what you do, don’t wait for an ecosystem to form around your startup. It won’t happen. Instead, start shoveling; if you’re lucky, you might find gold.
