AI businesses face a TAM trap as models keep improving
This week’s founder reading covers a structural risk baked into most AI businesses, a $240,000 API bill that nobody budgeted for, a gross margin problem with no clean solution, and a company that went from $0 to $1B ARR in two years.
Headlines and Trends
If better models shrink your market, you have a timer problem
The framing here is worth bookmarking. Most AI businesses operate in the gap between what models can currently do and what customers actually need. Every new model release eats a layer of that gap. If a model getting smarter is bad news for your product, you’re arbitraging a capability ceiling that keeps moving. The question to pressure-test: does your TAM go up when models improve, or does it quietly compress?
Agents need hard budget caps by default
Agents reduce the friction of spinning up code and API calls, which means they also reduce the friction of running up a bill overnight. The argument here is that hard budget caps should ship as the default setting, not an opt-in. Businesses would rather see an error than a surprise five-figure invoice. People who want unlimited agent runs should have to choose that explicitly.
Salesforce and Atlassian are now charging for agent API access
SaaStr received an estimate of up to $240,000 a year for agent API access that had previously been bundled with its software subscriptions. Its revenue agent makes 35,000 to 40,000 API calls a day across its tools. The agent itself proposed copying the data into a $5 Postgres instance to reduce those calls, though SaaStr reports it hasn’t built that workaround yet. Salesforce is introducing access charges. Atlassian already charges. HubSpot’s fees, as of this writing, appear limited to its own agents, with third-party agents still uncharged in SaaStr’s account.
Strategies and Tactics
Google is signaling a major update with documentation rewrites
Google rewrote its guidance on AI content this week, and the pattern matches what it has done before major algorithm updates. Manual fact-checking and review of AI-generated content is now described as critical. Fake authors are now categorized as a form of deception. If your site publishes AI-generated content, the window to fix quality issues is before the update drops, not after.
Harvey hit negative 50% gross margins as usage grew under flat seat pricing
The setup: your AI product gets better, customers use it more, and each account becomes less profitable. Harvey reportedly hit negative 50% gross margins in June as usage scaled under flat seat pricing. The comparison with Legora identifies three exits from that trap: charge for usage, reduce the cost per request, or cap what’s included. One proposed benchmark for viability is whether customers would stay when the product retains 40 cents of each dollar after covering the cost of serving them.
Miscellaneous
Cognition went from $1M to $1B ARR in 24 months
Cognition crossed $1 billion in annualized revenue on September 25. It was at $1 million just 24 months earlier. The company had no customers and no revenue when its launch video went viral in March 2024. Since then it has raised three rounds in twelve months, the last at a $48 billion valuation, and acquired Windsurf, a company five times its own size. The company’s agent now writes roughly 90% of its own code.
Quick links worth a look
- A founder reports reaching $120K ARR by DMing people who had already complained publicly about the problem his product solves. His first 100 users came from direct conversations.
- Audiohook acquired Podscan after building its own product on top of Podscan’s podcast data.
- One essay argues AI agents could become the primary gateway to commerce and online activity, controlling which businesses receive traffic and spend.
I’m founder of BizStack at Cagri Sarigoz LLC and a passionate advocate for entrepreneurs.
With over 14 years in tech, marketing, and AI, including my role as Head of SEO at CitizenShipper and co-founder of TaleBot at Intale AI, I’m dedicated to sharing genuine, useful product insights and tips.
At BizStack, I aim to cut through the digital noise to provide clear, actionable advice.
And more than all else, I’m a father to a (always) little girl and a husband.
Contact me at [email protected] for assistance.
