Enterprise Software Watch: The Agentic Software Shift: The Transition From 'Software as a Tool' To 'Software as a Worker'
AI Agents Reshape the Enterprise Software Stack
The Agentic Shift: From Applications Employees Use to Platforms That Perform the Work Themselves
Enterprise software is undergoing its most consequential transformation since the shift from on-premises licenses to cloud subscriptions. The question is no longer whether AI will augment enterprise applications. It is whether AI agents will effectively replace the application as the primary interface—and, in many cases, perform the work itself.
In 2026, every major enterprise software vendor has staked its future on agentic AI. SAP has unveiled the "Autonomous Enterprise" with more than 200 specialized agents. Oracle has launched 22 Fusion Agentic Applications spanning finance, HR, supply chain, and customer experience. Salesforce is delivering over 50 out-of-the-box AI agents in Slack and its IT Service Desk. ServiceNow has opened its full system of action to any AI agent through Action Fabric. Microsoft has made Agent Mode the default experience across Word, Excel, and PowerPoint. The shift is not incremental. It is architectural.
Enterprise Software Market This Week
The underlying economics of enterprise software are being rewritten. The per-seat SaaS model—the industry's dominant commercial structure for two decades—is under serious pressure. When AI agents replace human roles, the number of seats in an organization shrinks while the value delivered by the software goes up.
Vendors are scrambling to adapt. HubSpot now charges $0.50 per resolved conversation. Salesforce has introduced Agentforce Flex Credits, billing $0.10 per standard action and $0.15 per voice action. ServiceNow has moved to a consumption-based model that bills for each AI skill execution. Salesforce CEO Marc Benioff has acknowledged the uncertainty, saying the company is following startups rather than leading the change.
The stakes are high. Gartner projects that 40% of enterprise SaaS will include outcome-based pricing components by the end of 2026. Forrester predicts that over 45% of organizations already use AI agents, with another 25% currently piloting them. IDC projects that by 2029, more than one billion AI agents will be operational globally, collectively processing more than 217 billion actions per day.
The Agentic Software Shift
The transition from "software as a tool" to "software as a worker" is the defining enterprise technology story of 2026.
Traditional enterprise software presented users with screens, forms, and dashboards. Employees navigated applications, entered data, and executed transactions. The software was a system of record—passive, reactive, and dependent on human input.
Agentic enterprise software inverts this relationship. Instead of users navigating applications, AI agents navigate the applications—and the data, workflows, and systems behind them—to deliver outcomes. Users describe what they want; agents figure out how to make it happen.
At Google Cloud Next '26, the company framed this as the "agentic era": "where AI agents proactively anticipate goals, reason through complexity, and orchestrate complex business processes". Google noted that some 75% of its cloud customers are already using its AI products, and more than 230 of them processed more than 1 trillion tokens each over the past year. Oracle's Steve Miranda captured the shift succinctly: "The way work gets done no longer matches the speed, complexity, or expectations of modern business. With Fusion Agentic Applications, we are moving enterprise software beyond passive systems of record and providing our customers with applications that can reason, decide, and act in pursuit of defined business objectives".
From SaaS Applications to AI-Native Platforms
The distinction between "AI-enabled" and "AI-native" is becoming critical.
AI-enabled applications add AI features—copilots, assistants, chatbots—to existing workflows. The underlying architecture remains unchanged.
AI-native platforms are built from the ground up around AI agents as the primary execution layer. The application interface becomes conversational. Workflows are orchestrated by agents, not navigated by humans.
SAP's Joule Work exemplifies the AI-native approach: "Instead of navigating individual applications and entering data across several screens, users will now interact primarily with Joule. By describing a desired business outcome, Joule will orchestrate the right combination of workflows, data, and agents to get it done". Forrester has noted that this makes "screen-by-screen ERP design obsolete over 36 months". Salesforce's Summer '26 Release takes a similar tack, with Multi-Agent Orchestration enabling "agents to work together as a unified team to solve complex, end-to-end workflows". ServiceNow has gone a step further, opening its entire platform to external agents through Action Fabric.
Major Vendor Moves
SAP: The Autonomous Enterprise
At SAP Sapphire 2026, the company unveiled its vision for the "Autonomous Enterprise"—a unified AI platform for building, contextualizing, and governing agents. The centerpiece is SAP Business AI Platform, which unifies SAP Business Technology Platform, SAP Business Data Cloud, and SAP Business AI into a single governed environment.
SAP is deploying more than 50 domain-specific Joule Assistants across finance, supply chain, procurement, HR, and customer experience, orchestrating a subset of over 200 specialized agents. The Autonomous Close Assistant can compress the financial close process from weeks to days. The company has made its agent runtime free through the end of 2026—"the most aggressive commercial move SAP has made,e" according to Forrester.
Oracle: Agentic Apps Embedded in ERP
Oracle launched 22 Fusion Agentic Applications across finance, HR, supply chain, and customer experience. The company's approach is "AI should execute work inside ERP". The applications are designed to "make and execute decisions within business processes by drawing on enterprise data, workflows, policies and approval structures already held in Oracle's systems". The company also expanded AI Agent Studio for Fusion Applications, adding an Agentic Applications Builder, workflow orchestration tools, and contextual memory.
Salesforce: From Seats to Outcomes
Salesforce's Summer '26 Release is focused on multi-agent orchestration and Slack-first workflows. The company is shifting Agentforce to outcome-based pricing, betting that AI agents will let it charge for results rather than seats. Agentforce's annual recurring revenue passed $1.5 billion, up more than 240%.
CEO Marc Benioff framed the goal as charging $2 for every $20 or $40 in revenue the software helps generate. The experiment will determine whether legacy enterprise software companies can reinvent themselves as AI agents reduce how often employees interact directly with applications.
ServiceNow: The Platform for Autonomous Work
ServiceNow has positioned itself as the "AI agent of agents"—connecting any model, any cloud, and any data source. Action Fabric opens ServiceNow's full workflow system to external AI agents. Project Arc, developed with NVIDIA, is a long-running, self-evolving autonomous desktop agent for knowledge workers. AI Control Tower provides governance across AI systems, models, datasets, and identities, covering more than 30 enterprise integrations.
Microsoft: Copilot Becomes the Operating Layer
Microsoft used Build 2026 to signal a fundamental shift: Copilot is no longer a chat window. Agent Mode is now the default experience across Word, Excel, and PowerPoint. The company unveiled Copilot Cowork, developed with Anthropic's technology and designed for long-running, multi-step work. Agent 365 provides a control plane for observing, governing, managing, and securing agents across an organization.
Microsoft shared striking adoption metrics: "tens of millions of agents have appeared in the Agent 365 Registry" during preview; the company has visibility into more than 500,000 agents inside Microsoft; and "agents have been generating more than 65,000 responses every day for employees". Paid Copilot seats are up more than 160% year over year, with daily active usage up tenfold.
CRM, ERP & Business Applications
The agentic shift is most pronounced in the largest enterprise software categories: CRM, ERP, and business applications.
In ERP, the competition has moved from cloud migration to AI agents that can actually run business processes. SAP and Oracle represent two distinct strategies: SAP focuses on orchestration across the enterprise; Oracle embeds AI directly into transactional workflows. Both face the same fundamental challenge: "Agentic AI only works if companies have clean master data, mature processes, governance, and strong integrations. And many enterprises still struggle with basic ERP standardization".
In CRM, Salesforce is leading with Agentforce, delivering specialized agents for sales, service, and marketing. In workflow automation, ServiceNow and Pegasystems are competing to become the central nervous system for enterprise AI.
AI Workflow Automation
SAP showcased its work with European energy giant RWE, using Industry AI to reduce unplanned downtime across offshore wind turbines. ServiceNow cited Honeywell, where AI assistant "Red" eliminated the majority of service desk conversations. The Home Depot is building on its partnership with Google Cloud, using Gemini Enterprise to power a suite of AI tools.
Data, Integration & Governance
As AI agents proliferate, data integration and governance have emerged as the critical success factors—and the primary points of failure.
Data integration remains the "long pole in the tent" for all AI initiatives. SAP's Knowledge Graph gives AI agents "a structured map of business entities, processes and relationships across a customer's SAP landscape". Google's Gemini Enterprise integrates models across the entire stack with "security at every layer".
Governance has become "the decade's control point". Yet governance is struggling to keep pace. Two-thirds (66%) of global CIOs are not fully confident that their organization's AI governance model can keep pace with the speed of deployment. IDC identifies lack of security and governance as the leading concern surrounding enterprise AI, cited by 48% of organizations.
SaaS Economics & Pricing
The economics of enterprise software are being upended. The per-seat model—which fueled two decades of SaaS growth—is losing its footing.
McKinsey's 2026 service operations data highlights the economic engine: an AI resolution costs $0.62, compared to $7.40 for a human agent. That gap is driving the shift to consumption-based and outcome-based pricing.
| Vendor | Pricing Model | Rate |
|---|---|---|
| HubSpot | Per resolved conversation | $0.50 |
| Salesforce | Per action (standard/voice) | $0.10 / $0.15 |
| Zendesk | Per automated resolution | $1.50–$2.00 |
| ServiceNow | Per skill execution | Variable |
| Sierra AI | Per resolved outcome | $1–$2.50 |
Industry analysts estimate that AI agent deployments will eliminate the need for 20–35% of enterprise SaaS seats by the end of 2027. Gartner warns that over 40% of agentic AI projects will be canceled by the end of 2027 due to escalating costs, unclear business value, and inadequate risk controls.
M&A & Competitive Landscape
The agentic AI shift is driving consolidation and new competitive dynamics.
Salesforce's $3.6 billion acquisition of Fin—a startup that charges only after tasks are finished—signals the premium placed on outcome-based business models. SAP's pending acquisition of Prior Labs for tabular workloads reflects the importance of non-SAP data integration.
The competitive landscape is bifurcating along model-alignment lines. SAP-Anthropic and Microsoft-OpenAI have become the two dominant alignments, leaving Salesforce, Workday, and Oracle as multi-model-neutral. This ecosystem fragmentation creates both opportunity and risk for enterprises choosing their AI partners.
What It Means for CIOs
For CIOs, the agentic AI shift presents a dual mandate: capture the efficiency gains while managing the risks.
Adoption is accelerating. A Logicalis survey found that 60% of global CIOs plan investment in agentic AI over the next year. A Mayfield survey found that 42% of organizations already have AI agents in production. However, only 21% of CIOs use Agentic AI today, with half still exploring or piloting use cases.
The governance gap is real. Two-thirds of CIOs believe employee training around AI risk management is inadequate; only 37% believe their own knowledge enables them to accurately assess the reliability of their AI providers.
The pricing model is in flux. CIOs must navigate a messy transition from per-seat to consumption-based and outcome-based pricing. As Forrester notes, "the efficiency gains they expect to keep for themselves are often transferred directly to the vendor through these new pricing structures".
The enterprise software industry is at an inflection point. The shift from applications employees operate to platforms that perform work autonomously is not a feature update—it is a fundamental rewiring of how software delivers value.
Three trends will define the next 18 months:
1. The application interface is dying. The screen-based, navigation-heavy enterprise application is becoming obsolete. Conversational interfaces—Joule Work, Copilot, Agentforce—are becoming the primary way users interact with enterprise systems.
2. Governance is the new competitive moat. The vendors that win will not necessarily be those with the smartest AI. They will be those enterprises trust enough to let AI make decisions autonomously.
3. The economics are being rewritten. The per-seat model that funded the SaaS era cannot survive an era where software replaces the humans in those seats. The transition to consumption-based and outcome-based pricing will be messy, but it is inevitable.
What to Watch Next
- Agent governance standards. Will vendors develop interoperable governance frameworks, or will agent management become a source of vendor lock-in?
- Pricing model evolution. Will outcome-based pricing become the new standard, or will hybrid models emerge?
- Ecosystem consolidation. Will the SAP-Anthropic and Microsoft-OpenAI alignments solidify, or will multimodel approaches prevail?
- CIO adoption acceleration. With 60% of CIOs planning agentic AI investment, how quickly will production deployments scale?
- Agent-to-agent interoperability. SAP has promised full agent-to-agent interoperability by Q4 2026. If successful, this could unlock a new layer of enterprise automation.
- The "agent washing" problem. Gartner notes that of thousands of vendors claiming agentic capabilities, only about 130 are genuinely agentic. Expect increased scrutiny of vendor claims.
The CODEW Stat
IDC projects that by 2029, more than one billion AI agents will be operational globally, collectively processing more than 217 billion actions per day.
Metadata: SEO Title: AI Agents Reshape the Enterprise Software Stack: 2026 Market Analysis | Slug: enterprise-software-watch-ai-agents-2026 | Labels: Enterprise Software, AI Agents, SaaS, SAP, Oracle, Salesforce, ServiceNow, Microsoft, Agentic AI
Reviewed by Erwin Castro
on
Wednesday, September 02, 2026
Rating: