Freshworks CTO urges enterprise AI governance as adoption scales

Freshworks CTO Murali Swaminathan spoke with SiliconANGLE's theCUBE about enterprise AI governance during a livestreamed interview, according to SiliconANGLE. Swaminathan said interoperability across AI systems and multilayered trust are essential for organizations attempting to scale AI, and he discussed the move from human-assisted workflows toward more autonomous operations. He described governance controls that Freshworks embeds at multiple levels of the stack, including LLM guardrails, agent-level boundaries to keep automation focused, and data anonymization plus sovereignty policies, per SiliconANGLE's report.

Industry-pattern observations: Enterprises pursuing production AI typically require interoperability across model endpoints, orchestration layers, and third-party tools to avoid brittle point solutions. Layered governance commonly surfaces as a practical approach: teams combine input validation, model output filters, audit logs, and regional data controls rather than relying on a single control plane. For practitioners, instrumenting telemetry and traceability at the model, agent, and data layers is a recurring engineering task when moving pilots into production.

Reporting frames governance and trust not as optional features but as functional prerequisites for delegating decisions to AI, especially where automation affects customer interactions or regulated data. The emphasis on data sovereignty reflects broader enterprise compliance needs around residency and privacy that influence architecture choices like regional deployments and anonymization pipelines. For product teams, embedding controls into the stack aligns with a shift toward providing governance primitives to customers rather than leaving them to assemble point solutions.

For practitioners: monitor how vendors expose governance primitives (for example, configurable guardrails, agent isolation, audit trails, and exportable provenance) and whether those primitives integrate with existing observability and security tooling. Observers should also track adoption signals: the availability of standardized interfaces for policy enforcement and the emergence of vendor-neutral interoperability layers that make it easier to combine models and agents from multiple providers.

"There's a lot of change that has happened in the AI world. There's so much proliferation of AI-native tools [and] third-party products," Swaminathan said. "[You need] trust in how you set up the AI, how you set up the controls, how it performs, how you can trace what's going on and then how we can report on what it did and what it did not do," he added, per SiliconANGLE.

What was reported are Swaminathan's observed priorities during the interview; SiliconANGLE's article provides the quoted material and summaries above. The report does not include customer metrics, implementation case studies, or independent evaluation of Freshworks' controls.