In this interview, Pete Warren asked Cohesity’s Rob Sadowski to unpack the company’s newly announced Cohesity Maestro and explained why an open, standards-based approach to AI is central to the future of data resilience. Maestro, Sadowski said, is built to integrate natively with the AI platforms organisations are already adopting — agents like Anthropic’s Claude, Google Gemini and ChatGPT — which teams increasingly use to build customised applications, dashboards and workflows around their own needs. Rather than resist that shift, Cohesity wants to embrace it, letting customers use the AI tooling of their choice while still tapping the full power of the Cohesity Data Cloud for security and resilience tasks. The guiding principle, repeated throughout, is to “let them work the way they want to work.”

Sadowski translated some of the jargon. MCP, he explained, is the standardised interface most platforms have adopted so agents can interact with them, exchange data and issue commands — sparing Cohesity from building bespoke integrations for every platform. “Headless” refers to interacting with a system programmatically through agents rather than through its user-interface “head.” Crucially, this doesn’t open the door to rogue AI: every action through that interface is governed by authentication, authorisation and role-based entitlements, so agents can only do what they are permitted to do. This openness, he noted, is consistent with Cohesity’s long history of open APIs, encryption-key standards and partner integrations — much as the internet itself is built on open source.

On resilience, Sadowski argued the concept has evolved. Where it once meant simply backup and recovery, the rise of ransomware and other cyber attacks means security must be “built in, not bolted on.” True resilience is the ability to keep a business running with minimal disruption whatever the threat — making data more flexible and more secure at once. Recovery, he stressed, is one of five stages of cyber resilience mapped to common frameworks; historically over-weighted toward protection and prevention, programmes now need far more focus, budget and people on response and recovery.

Regulation and sovereignty featured strongly. Heavily regulated sectors such as finance and critical infrastructure must prove they can withstand and recover from malicious events, while sovereignty adds a further dimension of genuine control over where data sits and how it operates. These mandates, Sadowski said, reflect how existential data has become — it is the fuel for AI, used to train and feed models, and the foundation on which innovation is built. Firms that can secure and govern it can safely unlock “dark data” they never previously analysed, using tools like Cohesity Gaia. He cited Estonia’s data-driven government as an example, noting that public trust came from proving data could be protected — “trust but verify,” demonstrated through rehearsed, tested recoveries.

Finally, addressing fears that AI will replace jobs, Sadowski framed it firmly as a “force multiplier,” not a replacement. In Cohesity’s own experience, AI makes the best coders dramatically more productive rather than redundant. Organisations using AI purely to cut headcount, he warned, miss the far greater opportunity: harnessing human creativity and extending people’s reach.