AI Agents Framed as New SaaS Execution Layer
Industry analysts are refuting the idea that "SaaS is dead," arguing instead that AI agents represent a new execution layer within existing SaaS platforms. This perspective suggests agents will expand the total addressable market and deepen customer lock-in rather than replace incumbents. Companies like Salesforce are already restructuring to embed agentic workflows directly into their core enterprise offerings.
- The global AI agents market was valued at approximately USD 7.63 billion in 2025 and is projected to reach over USD 182 billion by 2033, growing at a compound annual growth rate (CAGR) of 49.6%. Another forecast estimates the market will grow from USD 7.84 billion in 2025 to USD 52.62 billion by 2030, a CAGR of 46.3%. This growth is driven by the demand for automation and advancements in natural language processing. - Agentic AI governance frameworks are emerging to manage the risks of autonomous systems, focusing on aspects like goal alignment, escalation protocols for runaway behavior, and interruptibility. These frameworks are critical as traditional AI governance, which focuses on static model validation, is insufficient for agents that take continuous, autonomous actions. Standards like the NIST AI Risk Management Framework and ISO/IEC 42001 are being adapted for agentic systems. - Key architectural patterns for agentic AI include designs for reasoning, tool use, planning, and multi-agent collaboration. Common workflow patterns involve task orchestration, sub-agent delegation, and event-based coordination to build scalable and auditable AI systems. These workflows can be categorized by their level of autonomy, ranging from augmented (AI assists humans) to fully autonomous (AI handles end-to-end execution). - For AI agents to effectively consume APIs, a shift in API design is required, moving from data-centric endpoints to goal-oriented interfaces. Best practices include ensuring write operations are idempotent to handle retries from autonomous agents, providing structured and machine-readable error responses, and designing for stateless requests that don't depend on server-side session state. - Enterprise adoption is accelerating, with a late 2025 McKinsey survey finding that 62% of organizations are at least experimenting with AI agents. However, significant trust and transparency issues remain, with 57% of AI leaders not fully trusting their agents' outputs and only 9% of enterprises having a mature AI governance framework. - Salesforce is integrating agent-to-agent (A2A) collaboration into its architecture, where a central "Orchestrator" agent interprets user intent and delegates tasks to specialized "Function Agents". This is part of their "Agentforce" platform, which aims to embed autonomous digital workers directly into sales, service, and marketing workflows. In early 2025, Salesforce's own deployment of these agents on its help channels handled over one million support requests with 93% accuracy. - Developer experience for AI agents requires more than just documentation; it benefits from context-rich developer portals with tutorials and guides to steer agent behavior. Because agents learn from examples, API documentation should include realistic production data rather than simple placeholders to ensure agents learn correct usage patterns. - The rise of agentic systems is expected to dramatically increase the transactional load on IT infrastructure. While current large-scale mobile systems handle around 10,000 transactions per second, the agentic era could see this volume increase by two orders of magnitude to approximately 1 million transactions per second.
