The news
SaaS pricing is meeting a new kind of customer: AI agents that call software through its API rather than through a paid human seat. SaaStr, the software-industry community and conference business, has published two accounts showing how that shift can turn routine renewals into build-versus-buy decisions.
In the first, SaaStr founder Jason Lemkin wrote that one vendor, which he did not name, gave the company an estimate of $240,000 for its agents' API access. Lemkin said SaaStr's main agent, which it calls 10K, makes 35,000 to 40,000 API calls a day across connected applications. He named Salesforce (CRM) and Atlassian (TEAM) as vendors charging for agent API access and said HubSpot (HUBS) is moving the same way, initially for its own first-party agents.
The agent's suggested alternative was blunt. According to Lemkin, 10K proposed mirroring the vendors' data into SaaStr's own PostgreSQL database, an instance he put at about $5 a month, to reduce dependence on paid vendor APIs. Lemkin said SaaStr runs with three people and more than 21 agents.
In the second account, Lemkin wrote that the same agent, which SaaStr describes as its AI vice president of marketing and revenue, recommended dropping Calendly and building a custom booking tool. He said it was built in about 20 minutes using Replit. The homegrown tool pulls in prospect data, routes meetings by which sponsors each team member already works with, and tracks prospects who leave without booking, features that depend on SaaStr's own data, which he said Calendly could not access.
Lemkin also stressed limits. SaaStr keeps a buy-by-default approach and still uses Salesforce and other vendors, he wrote, building only for narrow workflows that depend on its own data. In the API post, he added that a company picking a stack from scratch might now avoid vendors that meter agent access.
The numbers
- Agent API access estimate from one vendor (SaaStr)
- $240,000
- Proposed Postgres alternative
- About $5 a month
- Daily API calls by SaaStr's main agent
- 35,000–40,000
- Median horizontal B2B multiple, H1 2026 (a16z via SaaStr)
- 2.7x revenue
- Unicorns with under 2 years of runway (a16z via SaaStr)
- 55%
Why CEOs should care
For CIOs and procurement teams, the lesson is about leverage. If an agent can copy a vendor's data into a cheap database, or rebuild a simple tool in minutes, the cost of walking away from that vendor drops. Before your next renewal, list which apps your agents mostly use through APIs rather than screens, and ask each vendor in writing how agent calls are metered, what API limits apply and whether bulk data export stays free.
For CFOs, new agent API fees can quietly replace the seat revenue vendors lose as agents do work people used to do. One heavy user's estimate came to $240,000. Model what your own agent traffic would cost under per-call pricing, and set a ceiling in contracts before usage grows.
Builders should weigh the hidden costs too. A mirrored database or homegrown booking tool needs security review, ongoing maintenance and an owner. CISOs should ask who patches it, who can see the copied customer data and whether copying it is allowed under the vendor's terms.
The bigger picture
Investors are already pricing this pressure. In SaaStr's summary of a16z's latest State of Markets report, horizontal B2B software traded at a median of 2.7 times revenue in the first half of 2026, against 9.1 times for infrastructure and AI and 4.6 times for vertical software. The same summary said 55% of US venture-backed unicorns have under two years of runway and about 25% are profitable.
General-purpose apps that agents use mainly as data stores look easiest to replace. Products tied to proprietary data, regulated workflows or deep industry knowledge look harder to swap out.
What’s next
Watch how Salesforce, Atlassian and HubSpot set and publish agent API pricing, and whether buyers push back at renewal. Bundled agent allowances, rather than metered fees, may become the main thing buyers negotiate.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
Companies in this story
