If your finance team has started flinching every time someone mentions “AI agents,” they have good reason. Agentic workflows consume tokens at 10 to 50 times the rate of simple chat interactions. A task that cost cents with ChatGPT now costs dollars when an autonomous agent runs it across multiple steps, tools, and iterations. That gap between AI demos and AI budgets is one of the biggest friction points slowing enterprise adoption right now.

Pegasystems thinks it has the answer. On July 14, 2026, the company launched Pega Infinity 26 and made a pointed promise: you will never pay a token tax.

What Pega Infinity 26 Actually Is

Pega Infinity 26 is what Pegasystems calls “the industry’s first agentic enterprise transformation platform.” It bundles agentic AI capabilities, built-in governance, and a pricing model built around outcomes rather than compute.

The core pricing shift: instead of charging per token consumed, Pega bills per resolved case. You agree on a price for the work being done, not for the model calls that happen to produce it.

This matters because enterprise AI pricing has quietly fractured over the past year. After GitHub, Anthropic, and OpenAI all moved toward consumption-based billing, heavy AI users started receiving invoices that bore no resemblance to what they had planned for. The move from flat subscriptions to metered usage created what Pega is calling the “AI token tax” — the surcharge that accumulates when agentic workflows consume inference at scale.

Pega’s architecture is designed to avoid it at the root. Their Predictable AI™ approach moves as much reasoning work as possible to design-time, so the runtime execution uses far fewer tokens to complete the same outcome. The company says this allows clients to “use as much AI as they need” without watching costs spiral.

Pega Infinity Studio

Alongside the pricing change, Pega shipped Pega Infinity Studio — an AI-native development environment for building enterprise applications. It integrates with external coding agents while embedding Pega’s own Blueprint AI to guide developers toward deployment-ready patterns. The pitch to IT leaders is faster delivery with less accumulated technical debt, since the environment enforces governance and compliance standards as part of the build process rather than as a downstream audit.

Why This Matters for Businesses Evaluating AI

The AI vendor market in mid-2026 has two pricing philosophies competing for enterprise budgets, and understanding the difference is not optional for anyone deploying agents at scale.

The first is consumption-based pricing. You pay for tokens, API calls, or compute time. Costs scale directly with usage. This is transparent at small scale and becomes unpredictable fast once agents are handling real operational volume.

The second is outcome-based pricing. You pay for work completed. Costs are predictable and align with the actual business value delivered. The provider absorbs the risk of inefficient model calls.

Pega is betting on outcome-based as the model enterprises will prefer once they have lived through a few surprise invoices. The risk is that it works only if Pega’s underlying architecture is efficient enough to make the math work at high volume.

For Gartner’s part, the firm put $234 billion in enterprise software spending at risk from agentic AI disruption — a figure that suggests the vendors who solve the cost predictability problem will have a significant edge.

The Governance Piece

One area where Pega is distinguishing itself beyond pricing is governance. Pega Infinity 26 combines agentic capability with what they describe as built-in governance that makes AI actions compliant, consistent, and auditable.

This is not a minor feature. Enterprises in regulated industries — finance, healthcare, insurance — cannot deploy AI agents that cannot be explained, audited, or reversed. Most AI platforms treat governance as a later consideration. Pega is building it as a foundational constraint, which means agents built on it should clear compliance review faster than those bolted together with external guardrails.

What This Means for Business

The token pricing problem is not going away. As AI becomes embedded in operations — running approvals, handling customer queries, generating internal reports — the compute costs compound. Any business that is deploying agents now, or planning to in the next 12 months, needs to understand their pricing exposure before it becomes a budget line that leadership questions.

Three things to consider as you evaluate AI vendors:

Ask about pricing at scale. What does your monthly invoice look like when 10 agents are running 500 tasks per day? Get the math in writing, not in demos.

Understand what governance looks like. For any workflow that touches customers, finances, or compliance, you need to know how the agent’s decisions can be audited and who is accountable when something goes wrong.

Think about the total cost of the outcome, not the cost of the model call. A cheaper per-token model that requires five times the calls to complete a task is not cheaper. Evaluate total cost per resolved case, which is exactly what Pega is now billing.

Enterprise DNA works with businesses across all of these decisions. If you’re building out an AI agent strategy and want a straight conversation about what costs actually look like at production scale, book a session with the Omni Advisory team.

The token tax is a real problem. The vendors who solve it will win significant enterprise deals in the second half of 2026. Pega’s approach is worth watching.