AI
Service is the New Software?
Alloy Newsletter:
Insights into the intersection of AI, Law, and Business.
As an attorney, your time is your stock in trade. Early in my legal career, a senior partner told me: "I wish we sold products instead of services. Then we could make money while we sleep." That line stuck with me:, and, I've spent years fighting an internal bias that products are inherently superior to services.
AI is challenging that assumption in a fundamental way.
Every time a frontier model ships a major release, a class of startups effectively disappears. The "build a thin wrapper" playbook is dead. But I don't think the answer is to out-product Anthropic or OpenAI. That's a race you can't win.
The companies that win over the next 4–5 years will be the ones that layer deep expertise and services on top of these models. Not competing with the models, but compounding off of them.
The smart people at Sequoia Capital published a piece today that captures this well: if you sell the tool, you're in a race against the model. But if you sell the work, every model improvement makes you faster, cheaper, and harder to displace. For every $1 spent on software, $6 are spent on services.
The real addressable market isn't software budgets...it's labor budgets. The senior partner wasn't wrong about the appeal of products. But the real opportunity right now is building a services business that runs like one.
