Skan AI Raises $63 Million for Workflow Context and AI Agents
Skan AI raised $63 million to expand screen-observed workflow models and agents, citing bank savings, rapid growth and serious employee surveillance risks.
Summary
On August 12, 2026, Skan AI raised $63 million in a Series C co-led by Cathay Innovation and Dell Technologies Capital, with Citi Ventures, Bloomberg Beta, State Farm Ventures and Wipro Ventures, taking total funding to roughly $120 million. Skan also launched Skan AI Blueprint and Skan AI Agents, joining Skan AI Intelligence in a closed-loop platform for discovering, modeling and automating work.
CEO Avinash Misra, Skan’s co-founder with Manish Garg, says frontier models lack context omitted by process documents, standard procedures and backend logs. Skan watches scoped desktop activity across spreadsheets, CRM, email and 40-year-old mainframes, then extracts intent and aggregates hundreds of workers into a context graph, capturing the 80% of work Misra says occurs between committed system states. Customers scope specific apps and URLs; data stays behind their firewall while only anonymized metadata reaches the cloud, an architecture approved by European works councils. Skan contrasts this with Celonis process mining, UiPath replay automation, and agents confined to ServiceNow, Salesforce or Microsoft.
Gartner research cited by Skan says only 8% of enterprises have agents in production and 95% of early deployments need redesign; an MIT study found 95% of generative AI pilots lacked measurable returns. Skan claims over $500 million in identified, not fully realized, customer value. One U.S. bank found $37 million in friction across 11.2 million context switches by 1,500 finance staff, then achieved $18 million in annualized savings, 32% lower transaction costs and 41% higher throughput; another runs 60% of anti-money-laundering cases with agents. Undisclosed revenue grew above 300% for a second year, net dollar retention is about 150%, and customers include seven of the ten largest U.S. banks, one-quarter of the Fortune 50, Unum and Mitie. Skan runs on Nvidia AI Enterprise and NIM microservices as demand grows for private appliances.
Positives
- Insurers using Skan typically gain roughly 25% productivity in core claims, while one customer doubled case volume without adding a claims specialist.
- One large bank checks live cases against a 600-page controls inventory, with agents alerting staff when required compliance steps are missed.
- Unum, the $13.8 billion benefits provider, and Mitie are public customers; Mitie executive Cijo Joseph credited Skan with accelerating its AI transformation.
- European works councils have approved deployments using scoped observation, on-premises data and anonymized cloud metadata.
- Skan’s discovery, intelligence and agent products create one closed loop as enterprises consolidate AI spending among fewer vendors.
Risks & concerns
- Reuters reported in June that Meta reduced mouse-click tracking after employee concerns, underscoring resistance to workplace observation technology.
- Misra acknowledged that insights from Skan have led some customers to reduce headcount in certain processes.
- Aggregated telemetry can still expose an underperforming team even when Skan does not identify individual employees.
- The claimed $500 million in customer value represents expected savings opportunities, only some of which customers have captured.
- Observed work can contain mistakes, shortcuts and inefficient habits, creating a risk that context models faithfully encode flawed processes.