For two decades, Workday's pitch to enterprises was simple: we are the system of record for your people and your money. Buy the seats, lock in the data, and stop fighting with PeopleSoft.
That model is now obsolete, and Workday knows it. Over the past 18 months, the company has spent close to $3 billion acquiring Sana, Paradox, Evisort, and HiredScore, brought co-founder Aneel Bhusri back as CEO in February 2026, and quietly rewritten the commercial contract that 11,500 enterprise customers operate under.
The new pitch, made explicit at Workday DevCon in late April: Workday is no longer just a system of record. It is the agent system of record — the registry, governance layer, and metered runtime that every AI agent touching HR, finance, or workforce data has to pass through. And the bill is now per-call, not per-seat.
This is the most consequential platform repositioning in enterprise SaaS since the original cloud transition. Every CIO with a Workday contract needs to understand what changed, what it costs, and what their agent strategy looks like 12 months from now.
What Actually Happened: The Anatomy of a Reinvention
The visible story is the acquisitions. The deeper story is the architecture.
The buying spree. Workday spent ~$3 billion on four targeted acquisitions, each filling a specific gap in an agentic platform:
- Sana ($1.1B, closed late 2025): conversational AI front-end and agent runtime, originally built as a learning platform
- Paradox: conversational recruiting agents, now leading the talent acquisition product line
- Evisort: contract intelligence and document AI, powering new document understanding across the platform
- HiredScore: AI-powered talent intelligence and matching
These were not opportunistic buys. Each acquired company brought a working agent product and a leadership team that now runs a major Workday business unit. Joel Hellermark (ex-Sana CEO) is SVP and AI General Manager. Adam Godson (ex-Paradox CEO) runs talent acquisition. This is the closest thing enterprise software has seen to a successful "acqui-platform" play.
The CEO swap. Aneel Bhusri stepped down as CEO in 2024 and handed the role to Carl Eschenbach. By his own account at the recent strategy summit, he came back because Workday had "lost its startup culture and the AI strategy was not clear." That is unusually honest for a public company, and the resulting changes have been aggressive.
The product reveal. On March 18, 2026, four months after the Sana close, Workday launched Sana from Workday — a conversational interface that replaces traditional menu navigation across the platform. Three components shipped simultaneously:
- Sana for Workday: chat-based interface for documentation, navigation, and workflow initiation, free to all customers via Flex Credits
- Sana Self-Service Agent: 300+ skills handling HR and finance support tickets. Already used by 400+ customers in early access, delivering documented results of 25% reduction in HR case volume and 20% increase in employee productivity
- Sana Enterprise: cross-system execution layer with 18 connectors at launch (Google Drive, Slack, ServiceNow, and more), available to HCM and Financial Management customers
The platform now operates on four pillars Workday calls Find, Act, Build, and Automate — knowledge retrieval, task execution, asset generation, and no-code workflow creation. As Hellermark put it: "Instead of dozens of tickets and handoffs, you ask for an outcome."
The governance layer. At DevCon in late April, Workday operationalized the Agent System of Record (ASOR), already in use by 1,200+ customers to register and observe AI agents. Alongside it shipped the Agent Gateway, which exposes Workday to external agents via the two emerging open standards: Model Context Protocol (MCP) and the newer Agent-to-Agent (A2A) protocol.
Fifteen launch partners signed onto the new Agent Partner Network: Accenture, Adobe, AWS, Auditoria.AI, Compa, Deloitte, Glean, Google Cloud, IBM, Kainos, KPMG, Microsoft, Paradox, PwC, and WorkBoardAI. Their agents now register through ASOR and are subject to Workday's authorization, audit, and policy controls.
The new commercial model. Per-seat licensing is being phased into a hybrid that uses Flex Credits — a metered consumption unit that ranges from 1 to 750 credits per action depending on complexity. Critically, direct API calls from external agents are now metered per-call, capturing revenue Workday previously left on the table when systems integrators or third parties piped data out for free.
Workday is providing complimentary annual Flex Credit allowances based on customer size to reduce the sticker shock and seed adoption. Public rate cards have not been published.
Why This Matters: The Architecture Beneath the Marketing
Strip away the announcements and there are three structural shifts every enterprise leader needs to internalize.
1. The system of record has become a control plane
For decades, "system of record" meant a database with workflows. The agentic version is something different: a registry and policy engine that answers four questions every time anything (human, agent, integration) touches sensitive workforce or financial data:
- Who is making this request? (identity)
- Are they allowed to? (authorization, role-based or otherwise)
- What data and actions are in scope? (privilege)
- What did they actually do? (audit trail)
Workday is betting that enterprises will not let arbitrary AI agents — Microsoft's, Google's, OpenAI's, an internal team's — touch HR, finance, or workforce data without these four questions answered first. ASOR is positioned as that gatekeeper.
The strategic implication: Workday gets to charge a toll on agent traffic flowing into and out of its data, regardless of who built the agent. That is a more durable revenue model than seat licenses, which were always at risk of being squeezed by AI-driven workforce reductions.
2. The pricing model is finally honest about value
The per-seat model rewarded headcount growth. The Flex Credit model rewards usage of valuable actions. A simple lookup costs almost nothing. A complex workflow that auto-categorizes 1,000 expense reports, validates them against policy, and pushes them to AP costs more — because it would have cost more in human labor too.
This pricing aligns Workday's revenue with customer ROI, which sounds good until you do the math on what an agent-heavy deployment looks like. Workday currently shows $400M+ in AI ARR with Q4 AI annual contract value of $100M+, growing at over 100% year over year. That is a small fraction of $9.9B+ in projected FY27 subscription revenue, but it is the fastest-growing line item by a wide margin.
The risk for customers: an agent-led enterprise running thousands of automated decisions per hour could see Flex Credit consumption scale faster than budget growth, especially if internal teams treat credits as "free" because they are bundled into the contract.
3. The integration story changed
Until DevCon, the question "how do I connect Salesforce to Workday?" had a familiar answer: API integration, point-to-point or via an iPaaS. Through the Agent Gateway, the answer is now: an external agent uses MCP or A2A to negotiate with Workday's agents, and the work happens conversationally across systems.
This is a meaningful architectural shift. The integration is no longer a static data pipe but an agent-mediated handshake, with Workday enforcing policy on every interaction. Sana Enterprise extends this further by giving Workday a foothold inside Slack, Google Drive, ServiceNow, and other adjacent systems — a textbook expansion strategy that mirrors what Microsoft pulled off with Teams.
What This Means for the CIO and CFO
If your enterprise runs Workday, the next 12 months involve three concrete decisions, each with budget and architectural implications.
Decision 1: Lock in your Flex Credit baseline now. The complimentary allowances are generous on paper, but agent-driven usage scales nonlinearly. Force your Workday account team to model 12 and 24-month consumption under a realistic agent deployment scenario. Get the credit-per-action rate sheet in writing, even if it's only an internal Workday document. Without a rate card, you cannot do TCO math on agent-heavy workflows.
Decision 2: Decide your stance on ASOR before your business units do. Lines of business are about to start spinning up agents from Salesforce, Microsoft, Google, and best-of-breed vendors that need to touch Workday data. If you do not have a written policy on which agents must register through ASOR — and what governance, audit, and identity controls they must satisfy — your security and compliance teams will be retrofitting that policy under pressure later this year. The fact that 1,200 customers are already registered should tell you the train has left the station.
Decision 3: Audit your systems integrator spend. Workday is explicitly targeting the SI premium. Accenture reportedly reorganized 800,000 employees in one week using the new deployment tools. The Deployment Agent reduces configuration changes from months to under a week. If your annual SI spend on Workday changes is in the seven figures, ask hard questions about what shifts in-house once Sana and the new tooling are live. Some of that spend will be redirected, not eliminated — but the redirection is an opportunity to reset scope.
A fourth, more strategic question lurks behind all three: how much of your workforce intelligence layer do you want owned by a single vendor? Workday's pitch is end-to-end. The counter-argument — championed by best-of-breed vendors and increasingly by hyperscalers offering their own agent runtimes — is that locking workforce, finance, and learning into one platform creates a new single point of failure when agents become the primary execution surface.
What This Means for Builders and AI Teams
If you are designing or deploying enterprise agents that need to touch HR, finance, or workforce data, the architectural picture just got clearer in some ways and harder in others.
The good: open protocols. MCP and A2A support means you do not need to build a Workday-specific connector to participate. If your agent already speaks MCP — and most serious agent frameworks now do — you can register through the Agent Gateway and inherit Workday's identity, authorization, and audit model. That is significantly less work than writing custom OAuth flows and reverse-engineering REST APIs.
The cost: per-call metering changes design choices. Pre-agent integrations could batch, cache, and amortize API calls across many users. Per-call metering punishes chatty agents and rewards careful design. Decide upfront which actions actually need to round-trip to Workday versus what your agent can resolve with cached state or local logic. Track Flex Credit consumption per workflow as a first-class observability metric, not an afterthought.
The strategic: ASOR is the new control point. If you are building agents for an enterprise that runs Workday, you have to assume your agent will need to register through ASOR for any production use case. That means your agent needs a stable identity, declarable scopes, audit-friendly logging, and a clear ownership model. Treat this as a new compliance surface area on par with SOC 2.
The opportunity: Sana Enterprise as a distribution channel. Sana's 18 connectors at launch will grow. If your product or data lives in one of the systems Workday wants to reach (CRM, ITSM, code, knowledge, comms), there is a real opportunity to be the agent that completes a Workday-initiated workflow inside your surface. The 15-partner launch list is heavy on consultancies and infrastructure vendors. There is room for product companies that own a workflow.
The Bigger Picture: Enterprise SaaS Is Reorganizing Around Agents
Workday's reinvention is not happening in isolation. Salesforce launched Agentforce Operations on April 29 with similar workflow-control-plane positioning. Microsoft, Google, and AWS all shipped enterprise agent platforms in the past quarter. Every vendor with a privileged seat at the enterprise data layer is racing to become the agent registry, governance, and metered runtime for their slice of the stack.
The shape of the post-SaaS enterprise is starting to clarify. It is not a flat marketplace of best-of-breed apps integrated through iPaaS. It is a small number of agent control planes — almost certainly anchored by today's incumbents — federated through open protocols (MCP, A2A) and metered by consumption. The companies that win this transition will be the ones that already own the data of record and can credibly enforce governance over the agents that touch it.
Workday entered this race holding two of the strongest cards in the deck: workforce data and financial data. The $3 billion in acquisitions, the CEO swap, and the per-call metering model are how it intends to play them. The stock has not yet rewarded the bet — shares are at five-year lows around $118, down ~57% from peak — but the AI ARR growth curve and the ASOR adoption numbers suggest the strategy is landing where it matters: with the customers.
For enterprise leaders, this means the agent strategy conversation can no longer happen separately from the Workday contract conversation. They are the same conversation now. The companies that have that meeting in Q2 will save themselves a much harder one in Q4.
Sources:
- The Reinvention of Workday: From System of Record to Platform of Agents — Josh Bersin
- Workday DevCon - Agent Gateway adds third-party AI agents — Diginomica
- Workday Bets Big on Sana: 300+ AI Skills — iTWire
- 4 Months After $1.1B Sana Buy, Workday Goes All-In on AI Agents — Reworked
- Workday Stock Hits Five-Year Low as $400M in AI ARR Goes Unnoticed — TIKR
- Introducing Workday Flex Credits — Workday
- Workday Flex Credits: What We Know So Far — Commit Consulting
Want to calculate your own AI ROI? Try our AI ROI Calculator — takes 60 seconds and shows projected savings, payback period, and 3-year ROI.
