We’ve all heard a lot about the potential impact artificial intelligence and agentic AI will have on the SaaS sector. But its conversion from SaaS to GaaS is a new wrinkle.
The acronym technically stands for “governance as a service.” But it was used in a different way by Nvidia CEO Jensen Huang during this week’s GTC 2026 keynote where he said, “Every SaaS company will become a GaaS company.” Accuracy aside, rather than software that enables employees to do work, Huang’s formulation describes software that does the work itself, autonomously, through AI agents executing tasks without continuous human input.
The conceptual distance between SaaS and GaaS is not simply a product upgrade. It is a redesign of enterprise software’s basic premise. SaaS was built for humans to log into, navigate and act upon. GaaS is built for agents to receive instructions, plan execution sequences, and deliver outcomes with minimal human intervention in the loop.
Bain has mapped this transition across a three-layer stack: systems of record at the base, agent operating systems in the middle and outcome interfaces at the top. Microsoft Azure AI Foundry, Google Vertex AI Agent Builder and Amazon Bedrock Agents represent early platforms for building and running AI agents rather than pure orchestration layers. They include capabilities such as task planning, tool use and workflow coordination within individual agent environments.
The orchestration layer is emerging above them as a distinct control plane that manages multi-agent coordination, governance and execution across enterprise systems. In an enterprise deployment, a single business process can require a procurement agent, a compliance agent and a finance agent to coordinate in sequence, each passing context and results to the next.
As PYMNTS reported, companies are rapidly building platforms to manage these agent interactions at scale, including Nvidia’s enterprise agent stack and OpenAI’s push into enterprise-grade agent services, both aimed at giving businesses centralized control over how agents are deployed, monitored and governed across workflows.
Advertisement: Scroll to Continue
Software Designed for Agents, Not Users
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
PYMNTS previously wrote that seat-based models become structurally misaligned when agents, not humans, are executing tasks. A shift toward outcome-based pricing like charging for tasks completed, tickets resolved, and workflows closed rather than for access credentials will happen. Salesforce and Intercom have already begun moving in this direction.
The redesign runs deeper than pricing. Fortune observed that the AI reset forces companies to ask a foundational question: if the company were founded today knowing what AI can do, how would it solve the customer’s problem? For many enterprise software vendors, the answer would not produce what they currently ship. Products, pricing and operating models built around human-operated interfaces are being pressure-tested against a world in which the primary user is an agent querying an API.
PYMNTS also reported that emerging marketplaces, such as those being developed by Anthropic and OpenAI, are positioning agents as modular services that enterprises can select and deploy based on task performance. This model allows companies to mix and match specialized agents across vendors, accelerating a shift away from all-in-one SaaS platforms toward composable, agent-driven ecosystem
CFOs Register High Expectations as Adoption Accelerates
PYMNTS Intelligence research on chief financial officer attitudes toward agentic AI provides a ground-level view of where enterprise expectations are concentrating. The study found 43% of CFOs expect high impact from agentic AI applied to dynamic budget reallocation, with another 47% expecting moderate impact. Taken together, 90% of finance executives anticipate meaningful operational change from autonomous agent deployment across their function.
The same research found that 70% of firms already use at least one AI tool for cash flow management. The impact of that existing deployment is measurable: AI has reduced cash flow unpredictability from 68% of firms reporting cash timing as a chronic uncertainty to 17%, a reduction that reflects the transition from reactive, spreadsheet-driven forecasting to continuous, agent-assisted monitoring.
Those figures carry weight in the context of Huang’s GTC framing. The CFO data represents the early, bounded deployment of AI in finance workflows, before autonomous agent execution is fully normalized. The move from AI-assisted cash flow management to an agent that independently recommends, schedules and executes capital movements is the GaaS model in practice.