Skan AI raised $63 million in Series C funding on August 12, 2026, arguing that the real blocker for enterprise AI is not model capability but a nearly invisible data problem: the gap between how work is documented and how it actually happens. The round was co-led by Cathay Innovation and Dell Technologies Capital, with participation from Citi Ventures, Bloomberg Beta, State Farm Ventures, and Wipro Ventures. Total funding for the seven-year-old company now stands at approximately $120 million.
The announcement landed alongside general availability of two new products, Skan AI Blueprint and Skan AI Agents, completing what the company describes as a full platform loop from workflow discovery through autonomous execution.
The Problem: Agents Are Working from a Fiction
Enterprise AI adoption has been running far ahead of enterprise AI results. Gartner research cited by Skan finds that only 8 percent of enterprises have AI agents in production. More damning, 95 percent of early implementations will require a complete redesign. Those numbers echo a widely cited MIT study that found roughly 95 percent of enterprise generative AI pilots delivered no measurable return.
The conventional diagnosis is that agents need better models. Skan AI’s co-founder and CEO Avinash Misra argues the diagnosis is wrong. “Everyone is obsessed with building a better driver,” Misra told VentureBeat. “We think the bigger opportunity is building a better navigation system.”
His core claim is that the source data most enterprises use to ground agents was never accurate. Process documentation describes how workflows should run. System logs record committed states after decisions are made. Neither captures the handoffs, exceptions, workarounds, and institutional habits that define how a compliance team or a claims department actually operates. An agent trained on that data is not working with an approximation of reality; it is working with a sanitized fiction.
This is precisely why enterprise AI agent adoption lags so far behind the deployment timelines enterprises predicted in early 2026. The infrastructure was there; the context was not.
How Skan Works: Observation Before Automation
Skan deploys a lightweight sensor on employee desktops that watches how work moves across applications in real time. It tracks application switching, workflow sequences, decision paths, and time allocation by process. The captured signal covers the full application landscape, including the 40-year-old mainframe, the unstructured email thread, and the spreadsheet that somehow became mission-critical.
Nothing leaves the employee’s machine as identifiable data. Individual identifiers are replaced with tokens at the point of capture. Only anonymized, statistical metadata reaches the cloud analytics layer, aggregated across hundreds of people performing the same process rather than tracking any individual. The company has passed works council review in European deployments, including an approved rollout at Allianz in Munich.
The result is what Skan calls a context graph of work: a living model of how a business actually runs, derived from direct observation rather than inference from documentation.
“Think of it this way: if I were to share my screen, and you were to observe it going from Excel to CRM to email, in about two iterations you would build a model of what I do,” Misra explained to VentureBeat. “Except you could not do that at scale, you could not do it 24/7, and for 1,500 people like me. Now replace yourself with our technology.”
The company also contrasts its approach with process mining vendors like Celonis, which reconstruct workflows from backend system logs. Skan’s position is that log-based reconstruction misses the 80 percent of work execution that happens between committed transaction states, inside the applications rather than across them.
The Three-Product Platform
The Series C launch brings together three products as an integrated cycle:
| Product | Function | Output |
|---|---|---|
| Skan AI Blueprint | Discovers and prioritizes AI opportunities across all systems, including legacy | Automation opportunity map, prioritized by value |
| Skan AI Intelligence | Benchmarks processes, identifies variants, surfaces optimization opportunities | Operational dashboards and workforce management data |
| Skan AI Agents | Executes autonomously using observed-work context, with continuous real-world updates | Agent Operating Procedures (AOPs), live audits |
Each product feeds the next. Blueprint finds where to automate; Intelligence provides the operational data layer; Agents execute against the context Blueprint and Intelligence produced. The cycle is continuous: as real work changes, the context graph updates, and agents retrain against reality rather than a snapshot taken at deployment.
Customer Evidence: A Bank Case Study
The most concrete data point in the announcement comes from a top US bank. Skan observed 11.2 million context switches across 1,500 finance professionals, identifying $37 million in operational friction across workflow variants. Agents built on those observations reduced cost per transaction by 32 percent, increased throughput by 41 percent, and delivered $18 million in annualized savings.
That example illustrates why financial services customers are Skan’s strongest cohort. Banks operate under compliance requirements that make documented process fidelity non-negotiable, and they run on application stacks that mix modern SaaS with decades-old core systems. Agents grounded in real observation can enforce compliance against how cases actually move, not against a 600-page policy document no human holds fully in memory.
NVIDIA’s Global Head of Banking, Aser Blanco, framed the enterprise value case in the announcement: “A financial institution’s real differentiation is not its products. It is the decades of experience and operational know-how that shape how things get done internally. Skan AI observes thousands of real cases to capture how the enterprise’s best performers operate, then turns that into agents that run the work their way.”
What Investors Are Signaling
The composition of the round reflects Skan’s strategic position. Dell Technologies Capital brings enterprise infrastructure distribution. Cathay Innovation brings European scale, important as regulated deployments in banking and insurance are a core growth market. Citi Ventures and State Farm Ventures are both strategic financial services investors, signaling that Skan’s largest vertical is deepening its commitment, not just buying in.
Cathay Innovation partner Simon Wu made the infrastructure comparison explicit: “Enterprise work context is becoming the foundational infrastructure layer for enterprise AI the same way CRM became the system of record for customer relationships.”
That framing matters for how enterprises should think about Skan AI in a competitive stack. CRM incumbents like Salesforce became durable because they sat at the intersection of data, process, and workflow across the enterprise. Skan’s bet is that work context, derived from direct observation, can occupy an analogous position in the agentic AI stack.
Recent data on enterprise agentic AI deployment from Salesforce’s 2026 Agentic Enterprise Index showed that agent ROI correlates strongly with the quality of context agents are given at deployment. Skan AI is building the infrastructure layer that closes that quality gap.
Implications for Enterprise AI Leaders
The Skan AI funding highlights a pattern that enterprise AI architects are beginning to recognize: the model is rarely the constraint. What fails is the operational foundation that agents are supposed to reason over. Better documentation does not fix a documentation problem; it produces better documentation of a still-inaccurate picture.
For enterprise leaders evaluating agentic AI deployments, Skan AI’s platform points toward a set of questions worth asking before any agent goes into production. What is the source of the context the agent is using? Was that context derived from how work is documented or from how work is actually observed? How will the context update as the business evolves?
The company’s 150 percent net dollar retention rate suggests customers who deploy the full platform are expanding, not pulling back. Revenue grew more than 300 percent year over year. Those metrics indicate that the context graph approach is producing outcomes that enterprises want to scale, rather than a compelling pitch that does not survive contact with production.
For teams building enterprise AI agent infrastructure, the Skan AI Series C is a signal that the market is developing a new infrastructure category: real-time work observation as the operational data layer that everything else runs on.
The two new products, Blueprint and Agents, are generally available as of August 12, 2026. Skan AI runs on NVIDIA AI Enterprise and NIM microservices. Enterprise inquiries are handled through skan.ai.