SaaS pricing enters new era with AI agents, caveats included

The software-as-a-service (SaaS) industry has seen few shifts as disruptive and transformative as the adoption of AI. Now, with autonomous and semi-autonomous AI agents that perform tasks with minimum to zero human intervention, the industry is witnessing a new inflection point. In the span of just 12 months, agentic AI has begun to redefine software delivery and pricing models.

Traditionally, SaaS pricing has often involved subscription-based models, where customers pay a recurring fee (monthly or annual) for access to the software and its features on a per-user basis. They need large capital expenditures upfront, with long-term commitments for bundled features. But with the advent of AI agents that often work independently or across different functions, the set pricing model faces foundational challenges with the conversation shifting to consumption and outcomes.

This has led to the emergence of newer models of SaaS pricing that also claim to account for AI agent implementation.

Take, for instance, conversational AI firm Gnani.ai, which has developed an agentic AI platform called Inya.ai, uses different models for offering its AI agents to customers.

“We offer flexibility to our customers through models like pay-per-usage, outcome-based (e.g., per loan disbursed), and agent-per-month (like our Agent Assist). Agentic AI allows for micro-level pricing, where you only pay for the specific capability or task the agent is built to perform. These models eliminate upfront costs and let customers scale gradually. Each agent functions like a mini SaaS product,” Ganesh Gopalan, Gnani.ai’s CEO, told TechCircle.

He further added that among these, usage-based pricing stands out as particularly favorable, especially as agents become more autonomous.

Gnani.ai’s case reflects a broader trend across the industry. According to market intelligence firm Gartner, by 2030, at least 40% of enterprise SaaS spend will shift toward similar usage, agent- or outcome-based pricing. Many industry experts feel that this is still a conservative estimate and that the percentage could be much higher.

That said, for the pricing model to standardise in the Agentic AI world would require time. “AI agents bring with them a great deal of unpredictability, especially in terms of pricing,” said Arun Chandrasekaran, Distinguished VP Analyst at Gartner. He is also one of the authors of the study quoted above.

It is not always clear who is actually using the agent. When agents are priced based on tasks, it requires a high degree of attention to micro-pricing.

Since agents often involve large language models and process a significant number of tokens, organizations may suddenly face unexpectedly high bills due to the extent of the background work the agents were doing. This makes cost management, visibility, and governance critical issues in task-based pricing models.

There are also social and organizational dynamics to consider. With agent-based pricing, there’s often resistance due to the perception that machines are replacing human jobs. This can lead to pushback from within organizations or even from customers who may be uncomfortable with the implications of automation. “It's important to recognize these social nuances alongside the financial ones,” Chandrasekaran added.

On the vendor side, these new pricing models bring a whole new challenge to the fore, especially for startups and emerging players. “Unlike traditional SaaS models with annual contracts that provide steady, forecastable income, newer pricing models can fluctuate significantly. This makes it harder for investors to evaluate or value companies with such pricing strategies,” Gnani.ai’s Gopalan said. However, he is optimistic that as the industry matures, investors are likely to develop new frameworks to assess such models.

What to expect next

“SaaS contracts will increasingly resemble cloud utility bills or outcome-based agreements, with dynamic measures/KPIs, performance SLAs, and gain-share clauses replacing flat-tiers commercial models. Providers that can meter in real time and prove ROI will strengthen their pricing position and earn customer loyalty, while those continuing to offer fixed subscriptions will risk margin erosion and competitive displacement,” said Abhinav Johri, Technology Consulting, Partner, EY India.

The responsibility will lie with the vendor to demonstrate the value of the agent. Unlike integrated ecosystems, most organizations would not adopt AI agents unless they show measurable gains, like reducing the need for human effort, increasing productivity, or speeding up task completion.

“In summary, AI agents will sit on top of the existing SaaS budget, not as a bundled feature, but as an additional cost that must be justified. Their adoption will depend on clearly demonstrated benefits,” said Biswajeet Mahapatra, Principal Analyst at Forrester.

As AI agent pricing models evolve, chief information officers who focus on clear business outcomes will be best positioned to drive long-term impact and sustainable returns.