THE QUIET SHIFT: SAAS > TAAS

‍

AI has not merely created excitement in enterprise software. It has created a fair amount of quiet anxiety in board rooms too. SaaS challenges are real. Investors have begun asking a question that would have sounded absurd a few years back: what happens to a seat-based software company when its own AI grows capable enough to eliminate the very people who buy the seats?

‍

That is the sort of question that keeps a SaaS CEO awake at night. We suspect the honest answer is rather more interesting than the cynical one.

‍

A RATIONAL PROBLEM

‍

For two decades, Software-as-a-Service (“SaaS”) served corporations well. Vendors built platforms, enterprises assigned seats, and revenue scaled with adoption. It was not an elegant model, though it proved a sound one, sound enough to fund an extraordinary run of innovation across every industry it touched, healthcare very much included.

‍

Then artificial intelligence arrived, and the dilemma above stopped being theoretical.

‍

Consider the vendor’s position for a moment. If an agent can credibly perform the work once done by three coders, or a dozen claims specialists, the platform has created real value, and the health system benefits enormously. But the vendor’s own revenue was built on the assumption that a certain number of seats would remain occupied. If AI hollows that number out too quickly, the very success of the AI begins to erode the business that built it.

‍

This has little to do with greed and rather more to do with a business model colliding with a technology it never anticipated. Any vendor built on subscription economics must solve for this, or it will not survive the transition it just helped create.

‍

So the interesting question is not why vendors are doing this, but what a rational, well-run software company actually does when its own product threatens its own pricing model.

‍

ENTER THE GOVERNED PERSONA

‍

One answer we keep hearing described, in conversations with those close to several platforms, is genuinely clever: license the agent itself as a persona.

‍

Every great household keeps a ledger, not merely of who attends the ball, but of the staff as well. Enterprise software has long kept a ledger of its own: licensed users, assigned identities, authorized participants. Artificial intelligence is not necessarily striking names from that ledger. More often, it is adding new ones.

‍

Give the agent credentials, access controls, an identity within the system, much as one would onboard a human employee, a “governed persona.” We are told that some organizations now provision more agent personas than they have actual employees. Whatever else this is, it is not a shortcut. Governing an autonomous agent with the rigor once reserved for human access, audit trails, credentialing, role-based permissions, is a genuine engineering problem, and solving it well is no small feat.

‍

What it also does, whether by design or by consequence, is preserve the seat-based architecture that subscription revenue was built upon, while layering a second, usage-based charge atop it for the intelligence the agent actually consumes.

‍

We offer this as an observed pattern rather than an audited statistic, having not yet seen it spelled out so plainly on any rate card. But the shape of it grows familiar enough that We suspect a fair number of readers will recognize it.

‍

QUIET CONVERGENCE IN THE MARKETPLACE

‍

A term has begun circulating in AI circles for what is emerging: Tokens-as-a-Service, or TaaS. Whether or not that name ultimately sticks, this author believes it accurately describes the business model now taking shape, pricing that preserves the subscription economics enterprises already understand while introducing a consumption layer that scales with how much intelligence the system actually uses.

‍

This is more or less what one would expect a maturing market to settle upon. Neither pure strategy would have sufficed on its own. Straight SaaS pricing was never going to survive once AI began doing the seat’s job, and straight consumption pricing would have alarmed CFOs who budget in fixed license lines rather than variable usage bills. A hybrid was the rational middle ground, and it appears to be where the market is heading of its own accord.

‍

TOKEN TAX: A DEFINITION

‍

We want to be precise in defining “token tax” because the phrase carries more bite than the intent behind it. It is not offered as an accusation, but as shorthand for the new economic layer AI introduces into enterprise software purchasing, the portion of an invoice that scales with usage rather than seats, and one most buyers have not yet learned to evaluate the way they evaluate a license fee.

‍

Call it a cost center rather than a conspiracy. It is a genuinely new one all the same, and most procurement processes were not built to interrogate it.

‍

WHY HEALTHCARE EXPERIENCES THIS TREND EARLY

‍

Healthcare did not invent this convergence, and it will not be the last industry to encounter it. But healthcare’s enormous administrative surface area, paired with its long comfort with per-seat licensing across RCM, EHR, and clinical systems, makes it an unusually clear window into how the hybrid model is taking shape. Where seat count once tracked volume, token consumption increasingly tracks complexity, and the two are converging upon the same invoice rather than one replacing the other.

‍

QUESTIONS FOR THE NEXT RENEWAL CYCLE

‍

Rather than asking a vendor which model powers their platform, We would suggest a different line of inquiry to properly understand the economics on offer:

‍

• How does this pricing model balance seat-based and consumption-based economics?
• Which parts of my invoice scale with usage, and which remain fixed regardless of volume?
• As inference costs continue to fall industry-wide, how does that get reflected in my pricing over time?
• How is agent access and identity governed, and what does that governance cost me?
• What does “successful automation” actually look like in this contract, and how is it measured?

‍

WINNERS & LOSERS

‍

We do not believe the winners of this next era will be decided by clarity alone, any more than they will be decided by the size of the model behind the curtain. Trust gets a CFO to sign the contract but only measurable return earns the renewal. Vendors who last will be the ones who can explain their hybrid pricing plainly enough to be believed, and then back it with enough stacked value, above and beyond the token tax, that the invoice justifies itself on results.

‍

Healthcare, we suspect, will show us the answer before most other industries even finish asking the question.

‍

===================================

‍

‍Larrian Martin is the Chief Information Officer at GoSB, a specialty healthcare RCM company, He was formerly Senior Vice President at Envision Healthcare and holds an Engineering Science degree from Penn State University, where he studied in the Artificial Hearts Lab.