About
We built Strattum because enterprise AI agents kept hitting the same data wall
The founders spent years building data integration systems for regulated companies: insurance carriers, mid-market financial services firms, healthcare organizations. When AI agents became viable tools for those same companies, the data access problem followed the same pattern every time: no audit trail, no access controls, custom connectors that broke when source schemas changed.
We founded Strattum in 2025 in New York to build the middleware layer that was missing. Not another data warehouse, not another pipeline tool: a governed context layer specifically designed for AI agents working in regulated environments where data access cannot be uncontrolled.
Team
The people building Strattum
Priyanka Desai
CEO and Co-Founder
Priyanka spent her earlier career building enterprise data integration platforms for regulated industries: compliance-heavy environments where every system of record had a different data contract and every audit was a scramble. She led data architecture at a mid-market financial services firm before founding Strattum. Her focus is on making governed context delivery practical for the engineering teams who have to implement it, not just the compliance teams who require it.
Marcus Osei
CTO and Co-Founder
Marcus built distributed systems and API gateway infrastructure at a large enterprise software company before turning his focus to high-throughput data pipeline tooling for analytics teams at two consecutive data infrastructure startups. He designed the Strattum connector protocol and the policy engine's evaluation architecture. His design principle for the platform: every component should fail loudly and informatively, never silently and ambiguously.
Tamar Feldman
Head of Engineering
Tamar's background is in compliance automation and data pipeline engineering. She spent several years building access-control systems for a regulated data environment in the healthcare sector, where access review cycles and data classification accuracy were constant engineering constraints. She joined Strattum to build the engineering team and the connector SDK from scratch, with a particular focus on making the audit event schema complete enough to be genuinely useful in a compliance program.
How we work
Three operating principles
Build for the audit, not after it
Compliance review requirements are known in advance. We build the audit trail structure to answer access-review questions from day one, not as a feature added when a customer needs to pass an audit.
Connectors that fail loudly, not silently
A connector that returns a partial result without surfacing the reason is worse than a connector that returns an error. When a source schema changes or a credential expires, the connector says so clearly before the agent acts on incomplete context.
Governed access grows trust faster than locked-down access shrinks risk
Blocking AI agents from enterprise data reduces risk in the short run but stalls the programs that regulated companies need to stay competitive. Governed access with full traceability unlocks the same programs without the compliance exposure.
Get in touch
Talk to the team
Whether you are evaluating connectors for a regulated industry use case, have a compliance architecture question, or want to join the design partner program: we read every message.