Today, there are few undisputed AI winners; companies have an urgent opportunity to nail their pricing strategy and claim category leadership. In this cutthroat race, getting it wrong will have company-ending consequences.

From outcome-based pricing to token fatigue, here are five key trends that emerged from our recent conversations with senior sales leaders across AI-native startups and incumbent platforms incorporating AI.

1. Token-based and pure per-seat pricing models are losing favor

Although token consumption models were once seen as the most natural way to price AI (especially for LLM-powered products), many vendors and buyers alike now find them unsustainable or opaque.

The continuous drop in token costs means that consumption-based pricing gets undercut with every model release. Savvy buyers ask to renegotiate their vendor contracts accordingly. The result for providers is a narrowing margin above variable costs, and an inability to cover fixed cost structures.

Pure per-seat pricing is similarly declining in popularity. While this model is easy for buyers to understand, it discourages the widespread adoption necessary for embeddedness and retention. A Sales leader at a conversational AI company underscored:

“Our pricing is focused on the problems that agents solve. It’s funny – a lot of our competitors used to be seat-based. We were the first ones to be like ‘No, it’s based on ticket volume. You can have unlimited users.’ Most of our competitors do that now or often.”

Founders should be wary of pegging pricing to a metric with downward trajectory – like model costs. While there are instances where per-seat pricing makes sense as a pricing component, it should not be at the expense of facilitating widespread uptake within a customer’s employee base.

2. Simplicity and predictability matter more than ever

Simplicity is generally considered a core pricing tenant. As Simon Kucher, cofounder and CEO of Index puts it: it should be straightforward to explain, scale with value, and be implementable while ensuring transparency and fairness in billing.

As AI’s footprint grows, so does buyers’ demand for pricing clarity. Several sales leaders emphasized the importance of predictability, especially when AI capabilities touch multiple parts of an organization.

For AI-native companies, that means making pricing as transparent as possible. A sales leader at an AI-enabled customer support company reflects on their decision to predominantly price per agent conversation (versus per resolution):

“Per conversation pricing is the easiest unit to predictably measure. Historical volume of conversations creates a very predictable model for what associated costs year over year will be. With outcome-based pricing, there can be questions on ‘how do you define a resolution?’”

For incumbents playing catch-up with their AI offerings, “simplicity and predictability” often means leaving pricing unchanged for existing logos. Competition is fierce; buyers view AI enablement as table stakes versus a line item that commands a price increase.

A VP of enterprise sales at a tech-enabled marketing company explained that while they’ve deeply embedded AI into their roadmap, they often don’t line-item it on quotes:

“It’s just part of the package now. And for new logos, we work out the AI cost with finance on the back end. This is going to help us attract and retain additional clients….Even if there may be a margin hit, the revenue and the downstream impacts of the deal are going to be worth it in the long run for us from a revenue growth standpoint.”

The consensus: complex, variable pricing is difficult to explain and harder to sell – especially to large enterprises with long procurement cycles. Buyers want to know how much they’ll be paying, and for what, in clear terms.

3. AI pricing is increasingly tied to ROI, not features

Demonstrating the clear ROI of software pricing was always difficult – and this is more the case now when software ROI can include reduced labor costs. If employees are partially or fully replaced with software, then that software can command a higher price versus prior SaaS tools. As a result of this complexity, some AI-native disruptors are using internal value consultants to quantify the ROI of AI. For example, at one company:

“We now have a team of value consultants who model cost savings and labor efficiency for customers over 2–3 years. It has helped reframe the conversation from ‘what are your model costs?’ to ‘how much are we saving you?’”

As is the case in any pricing proposal, the key is getting the customer to buy into the ROI that you're pitching. For the newest AI technology – where prospects have fewer experiences with value delivered – thoughtful POCs are used to prove impact.

A VP of strategic sales at a full-stack enterprise AI platform shared:

“We do a lot of POCs, a lot of customized demos and validation to get to that point where the prospect can say, ‘Yeah, we do agree that this is going to be valuable for us and going to give us a significant ROI.’”

Every sales leader we spoke with emphasized that POCs must be tightly controlled. Clear success metrics (e.g., resolution rates, humanless touchpoints) and enterprise minimums are defined before POCs begin, or the prospect is DQ’d. One company goes so far as to have four different POC archetypes it deploys depending on the prospect profile. This rigor enables the vendor to maintain control over the prospect's experience and better showcase the product's value in the context of the customer's specific needs.

4. Sales team feedback is driving rapid pricing model iteration

As a general rule, pricing should be approached flexibly as the market gives feedback. That’s even more the case now with AI changing software commercialization so rapidly. Several salespeople at leading startups described highly fluid internal feedback loops where GTM teams directly shape pricing strategy.

Michael Weber, Head of Mid-Market Sales at Ramp shared how tightly aligned Product and Sales are:

“We optimized pricing for [our new product] based entirely on AE feedback. There is deep respect between Product and GTM leadership.”

Others echoed the importance of deal-desk agility:

“We recently revamped our pricing. Before rolling it out, the team collected feedback from reps, listened to Gong calls, and shaped the new structure accordingly.”

This iterative approach isn’t just healthy; it’s essential in an environment where AI product development and usage patterns evolve almost daily. Founders: your Sales folks are the voice of the customer. Establish clear feedback mechanisms to ensure you are constantly iterating pricing based on patterns they surface – particularly from failed deals.

5. Successful AI selling still relies on tried-and-true sales methodology

Perhaps to the relief of folks reading this, a lot of what fuels a successful deal is still anchored in being a great salesperson (think: MEDDPICC).

A serial builder of enterprise sales teams shared their reflections:

“The more I've thought about it, the more I'm like, ‘this is kind of the same as everything else’. The adoption curve is a little bit different. But, when I think about the dynamics of the deals that we've sold, most of them look like other big enterprise strategic deals.”

Jacquelyn Goldberg, VP Sales at Unframe AI seconded this sentiment, with one addition:

“I 100% agree with that, except what I would add is in this new world of AI, especially within the startup ecosystem, you need very creative sellers. Not ones that just plug into a process. I need folks that are curious about AI so that they can do better discovery and fully understand how it works.”

If your sales team is overwhelmed, go back to basics: ask good questions, really listen, and frame outcomes in simple language your prospect understands.

Pricing in a post-hype AI market

From per-agent fees to platform subscriptions to “percentage of impact” pricing, there’s no single right way to price AI. What’s clear is that buyers expect transparency, sellers need flexibility and everyone’s figuring it out in real-time.

If there’s one universal takeaway from these conversations, it’s this: AI pricing is not about the technology itself. It’s about how clearly a company can define (and prove!) the value it creates.

As one sales leader put it: “Our goal isn’t to sell AI. It’s to sell outcomes. AI just happens to be how we get there.”

If this resonated, or you’re currently rethinking your own pricing model, I’d love to hear how you're approaching it. Reach out at ncoletta@baincapital.com!