Repello AI raises $1.2M seed to build red‑teaming and safety guardrails

Repello sits on the defensive side of the AGI race, which is finally starting to attract dedicated startups rather than being tacked onto broader security suites. As enterprises move from experimenting with LLMs to wiring them into workflows and customer experiences, attack surfaces like prompt injection, prompt leakage and output manipulation become business-critical. Building tools that can automatically red-team systems and enforce guardrails at runtime is a prerequisite if we want widespread deployment of increasingly capable models without constant “jailbreak of the week” headlines.

The Bengaluru–San Francisco footprint is notable: India brings a deep bench of security and infra talent plus proximity to fast-growing AI adopters, while the Bay Area connection plugs into the frontier-model ecosystem. This dual presence could make Repello an attractive integration partner for both Indian AI champions and US labs that don’t want to build full-stack safety tooling themselves.

Zooming out, early funding rounds like this suggest that security for generative and agentic AI is crystallising into its own product category, akin to the rise of cloud security a decade ago. The better these tools get, the easier it becomes for mainstream enterprises to justify more aggressive use of near‑AGI capabilities without running afoul of compliance or reputational risk.