Both deals landed on Columbus Day. On October 12, a holiday when much of enterprise tech was offline, Workday signed to acquire C1.ai and Freshworks signed to acquire Torii. Neither disclosed price. Both expect to close in Q4. In both cases an enterprise platform company is buying into agent governance.
They picked different control points. Workday is buying the authorization layer, the system that decides whether an agent can act before it acts. Freshworks is buying the accounting layer, which tracks what agents consumed, who consumed it, and what it cost. Freshworks–Torii gets its own companion piece. This one covers Workday–C1.ai.
What C1.ai Actually Does
C1.ai, formerly ConductorOne, began as a human identity-governance product. It connects HR systems and directories to application connectors, then uses employment events (a hire, a role change, a departure) to trigger access workflows: access reviews, provisioning, separation-of-duties checks. Standard identity governance, done well enough to land Brex, DoorDash, Instacart, Ramp, and Zscaler as customers.
For this deal, what matters is what C1 shipped in 2026: three capabilities built for AI agents that need governed access to enterprise systems.
Agent identity management. C1 treats each agent as a governed service principal, meaning the agent gets its own identity with an owner, approvers, entitlements, and a review schedule. Most organizations currently do something else. An engineer with broad system access builds an agent, and the agent quietly inherits that engineer's access with no independent review.
Shadow AI discovery. C1's discovery product scans two surfaces. Cloud connectors find unowned agents and map the MCP servers, APIs, and data stores those agents can reach, flagging long-lived keys and over-permissioned service accounts. (MCP, the Model Context Protocol, is the standard way agents connect to tools and data.) Endpoint scanning looks at developer machines for local MCP configurations, unsanctioned copilots, and plaintext API tokens sitting in .env files. The operational problem is straightforward: a developer can stand up an MCP server without IT knowing it exists, and each one is an ungoverned path into enterprise systems.
Runtime tool-call enforcement. This is the most technically distinctive piece. C1's July 2026 release puts an identity-aware gateway between an agent and its tools. Every call routed through the gateway is checked against policy before it executes, with four possible outcomes: allow, block, hold for human approval, or redact designated fields from the returned result. What gets enforced is the credential, the destination, and the policy. The gateway does not rely on a model-generated judgment about what the agent "intended." If you've built agent systems, you know why that distinction matters. Intent is unreliable as a governance primitive; the authorization check is not.
C1 also built a credential-vending system that mints short-lived, scoped credentials on demand instead of embedding durable secrets inside agent workloads. The agent never holds the underlying secret, and revocation takes effect on the next call.
The Gap This Fills
Workday was not starting from zero. Its Agent System of Record, announced in February 2025, already gave customers a central agent registry covering onboarding, role definition, activation, monitoring, cost budgeting, and retirement. Each agent gets its own system account, and Workday's security framework applies the same domain and business-process policies it uses for human workers. Workday had also acquired Flowise, Sana, and Pipedream for agent building, enterprise search, and third-party application connectors. More than 400 customers were using role-based agents.
Those controls were deepest inside Workday, though, governing how agents interact with Workday data and Workday APIs. Enterprise agents don't stay inside. They reach into CRM, code repositories, cloud infrastructure, communication tools, and whatever MCP server a developer spun up last Tuesday.
C1's connectors, discovery tools, credential vending, and runtime gateway all operate across those external systems. Workday already owns the employee record, the organizational hierarchy, and the role structure, and it already built the agent registry. What it lacked was enforcement that follows the agent once it leaves Workday. C1 supplies that.
If you read the SAP–TechWolf and Salesforce–Listen Labs analysis, you'll recognize the pattern: an enterprise platform acquires external context its agents can't generate internally. SAP bought organizational-skill context. Salesforce bought behavioral context. Workday is buying cross-system authorization context — the ability to govern agent identity and tool access across every application an agent touches, including the many Workday doesn't own.
Why Authorization Might Be the Foundational Layer
Workday's bet has a structural logic beyond filling a product gap. Authorization sits architecturally upstream of every other agent-governance function.
The dependencies run in one direction. Attributing AI costs to an agent requires having identified the agent. Enforcing a spending limit requires the layer that controls execution. Auditing what an agent did requires a governed identity record for it. Discovery, cost tracking, compliance reporting, and performance monitoring all wait on the same prior questions: which agents exist, who owns them, and what are they allowed to do?
C1's product stack lines up with that chain. Shadow discovery answers whether an agent exists, agent identity management answers who owns it, and runtime governance plus credential vending answer what it's allowed to do. Once those are settled, cost attribution, compliance evidence, and performance analysis become tractable. Without them, you're reading aggregate invoices and hoping the numbers make sense.
The counterargument deserves weight. Most enterprises aren't ready for a mature authorization framework. They don't have agent registries, and they don't have policy engines governing tool calls. Building that infrastructure takes organizational commitment (policy design, connector deployment, security team buy-in), and the cost problem is already acute. A CFO who can't tell the board how much the company spends on AI needs an answer now, and consumption tracking can provide one without waiting for the identity layer to mature.
Workday is betting that the identity layer is where institutional knowledge accumulates. Policies, exceptions, approval chains, and review histories compound over time and get harder to replicate or migrate. A cost dashboard can be rebuilt on a different platform. An authorization graph holding three years of policy decisions, exception records, and audit evidence is much harder to move. If Workday is right, the authorization surface becomes the governance primitive that other layers, cost tracking included, depend on and plug into.
Whether Workday can actually build that surface by folding C1 into its platform is a separate question.
What Remains Unproven
Integration architecture is unpublished. There is no public design showing how C1's identity objects, policies, audit records, and gateway will map into Workday's Agent System of Record. Workday's SEC filing lists successful integration as a transaction risk. This is the question that determines deal value: does the deal produce a unified governance surface, or does C1 end up as a co-branded tool bolted onto Workday's platform? I raised the same integration question for SAP–TechWolf, and it applies here with equal force.
Runtime enforcement coverage depends on adoption. C1's tool-call governance covers only agents whose calls route through its gateway. Discovery can find an agent; enforcement requires that agent's traffic to actually flow through C1. The gap between what C1 can see and what it can control narrows only as customers deploy the gateway broadly.
Production evidence is thin. C1's AI Access Management became generally available in May 2026. The runtime-governance announcement disclosed no customer counts, production volume, false-positive rates, or latency data.
Maturity
C1 had raised approximately $111 million, including a $79 million Series B in October 2025 at a reported $350 million valuation. Forbes reported that revenue had tripled and that management was targeting $50 million in annualized revenue within two years. C1 has not disclosed current ARR or total customer count. Headcount was approximately 139 employees as of March 2026; treat that as an estimate, since Revelio Labs builds it from public profiles.
Workday is buying technology, a team, and a thesis — not a revenue engine.
