As Seat-Based Pricing Fades, These AI Startups Are Finding Alternatives

Pricing is a key question facing AI startups right now. The model almost every software company relied on for the last decade is breaking down in real time.

Seat-based pricing, the backbone of the SaaS era, was built for a world where software was assumed to be used by a client’s human employees – and the number of employees would grow as the client grew. As AI agents replace and augment human labor at many client firms, we believe continuous employee growth is no longer a guarantee.

“All the developer tools are becoming Claude Code or a command-line interface, so these days you tell them what to do, and you have fewer developers, fewer product managers and more agents,” says Manny Medina, founder of AI monetization platform Paid. “With agents in the mix, the seat model just breaks.”

The rise of new pricing models

For all its ubiquity, seat-based pricing had its flaws. It frustrated those customers who found themselves paying for sprawling bundles of software they barely used. AI is providing the trigger for experimentation with new pricing models, with no clear consensus appearing as yet. The dilemma is being debated at the highest levels.

“The topic of pricing is consistently bubbling up to the board, especially in the AI era,” says Davis Geidt, research practice leader at the Alexander Group, a go-to-market consulting firm.

So far, the software industry's search for new pricing models has been a process of trial and error. The first response was a swing toward consumption-based models where clients pay for processing each unit of AI data, known as a token. But that introduced a new problem: unpredictable bills and no clear link between spend and value delivered.

Now a third model is becoming increasingly common: outcome-based pricing, where clients pay for specific, measurable results. The appeal is obvious: costs are tied directly to business outcomes, such as qualified leads generated or customer support tickets resolved. This can better align vendor and customer incentives.

There’s one difficulty with this model. As go-to-market leaders are discovering, business outcomes are not always easy to define or measure. “It’s not without its challenges,” says Latané Conant, chief marketing officer at agentic AI customer service startup Parloa.

A new SaaS playbook

The seats model has endured as long as it has for one reason: predictability. Software businesses and their clients can easily budget and plan with seat-based pricing, while outcome-based pricing asks clients to give up that predictability. Not every client is ready to make that leap.

“Outcome-based pricing is likely the future, but you have to match what people are ready for,” says Katie Burke, chief operating officer at Harvey, an AI startup for the legal industry. “Customers don’t mind paying for value, but also need transparency in what they are getting and how to explain it to their finance teams.”

Many startups will likely take pages from the SaaS playbook and find some type of predictability wrapper, such as minimums or thresholds, to get clients comfortable with the pricing, while also ensuring sustainable revenue flows for themselves. Paid’s Medina says pricing is not only a conversation for a startup’s board; it’s also a conversation with clients, who will likely need to adapt to a changing strategy.

What’s working and what isn’t

Outcome-based pricing works well in use cases like customer support, where processes are defined and success is measurable. But even here, the model isn’t perfect. For vendors, the risk runs in both directions: unconstrained support conversations can consume tokens and massively increase costs; meanwhile, a dissatisfied customer could mean the outcome criteria are never met, and the work potentially goes unpaid.

“The AI could do a lot of work in a customer support interaction and still get no resolution,” says Des Traynor, co-founder of AI customer agent Fin, who has led from the front in the move to outcome-based pricing.

Then there is the problem of clients gaming the system. “Customers could potentially get value, but still structure their interactions so they never formally trigger the outcome criteria and don’t have to pay,” says Parloa’s Conant.

Some startups are exploring consumption-based pricing models, where costs are based on technology usage. This is similar to how hyperscalers like Amazon price their cloud services or how AI providers like OpenAI charge per token. However, in many industries – such as law, procurement or the sciences – consumption-based charging would be a big change and it could lead to customers underusing the product.

The pragmatic path

The AI startups navigating this best aren’t just chasing the newest pricing model because it sounds innovative, they’re asking a harder question: what structure actually reflects the value we deliver, and what are our customers ready for today?

The honest answer for most startups is somewhere in the murky middle of the pricing models we’ve discussed. Seats are predictable but are becoming increasingly hard to defend. The pragmatic approach is building toward outcome-based models while giving customers the familiarity they need to go with the company on that journey.

“There is a role for seats if you can make it an enabler of your growth,” Medina says. “But in most cases, the seats model is not going to be the leverage of growth; it’s just going to be a bridge between where you are right now and the move toward newer pricing models.”