Runtime Data Control for Enterprise AI Search
Wednesday, August 26, 2026
Skyflow, the runtime data control platform for agentic AI, announced Skyflow for Glean - a data security and governance layer for enterprise context. Enterprise context is the millions of documents and billions of records spread across systems that were never built to work together. Skyflow already protects billions of sensitive records for Fortune 500 companies. This launch extends that protection to the knowledge that powers enterprise AI search, starting with Glean deployments.
Securing Enterprise Context in Production
Kearney, a management consulting company, ran into this challenge directly. Rolling out Glean across thousands of consultants meant connecting terabytes of client work, all carrying strict confidentiality commitments to each client. Skyflow governs that context so consultants see only the engagements they're authorized for, with a full audit trail. Contd
"We don't want to be in the data sanitisation business," said Mark Johnson, Partner and Chief Information and Digital Officer at Kearney. "Skyflow gave us per-field encryption, runtime policies and customer-specific isolation as a platform – not a project."
Extending Governance to the Context Layer
AI agents don't work from one fixed dataset. They assemble context on demand from CRMs, SaaS apps, data lakes and wikis — so the real question is no longer who can access a system, but what data an agent is allowed to see. The usual fix is to block the riskiest content, but blocking degrades search: identifiers stop matching across documents, and employees end up searching a fraction of what the company knows.
Enterprise search platforms like Glean enforce permissions-aware access, ensuring employees and agents retrieve only the documents they're authorised to see. Skyflow adds a complementary layer of fine-grained control at the data field level — valuable because AI search turns raw documents into generated answers, summaries, and agent reasoning steps that can carry sensitive values. Skyflow for Glean works at two points:Sanitise on ingest - Skyflow detects and tokenises sensitive values before content enters the search index and embeddings, so raw values never reach the index or model memory.
Control the results - At retrieval and throughout agent execution, Skyflow enforces policy-based masking, rehydration, and context filtering, so users and agents see only what they're authorised to see.
"Every search and every agent action touches your most sensitive records," said Anshu Sharma, co-founder and CEO of Skyflow. "Protecting that moment matters as much as protecting the storage behind it. Security and legal teams can move from blocking access to unblocking secure access — and that's what makes AI search safe to turn on."
What This Means for Enterprise AI
With governance satisfied at the data layer, pilots become firmwide deployments. Employees search the complete knowledge base, one client's information never surfaces in another's results, and sensitive data stays out of answers and summaries that shouldn't include it. Data stays in the region for GDPR, DPDP, and HIPAA, and every access is recorded.
