Repello AI Raises $1.2M to Enhance AI Security with Automated Red Teaming

Repello AI Secures $1.2M to Automate Red Teaming and AI Guardrails | AIM

"We're at an inflection point where AI adoption is accelerating faster than security solutions can keep pace". Repello AI, a GenAI-focused security startup, has raised $1.225 million in seed funding to bolster its mission of securing generative AI systems against emerging threats. Founded in 2024, the company is building continuous AI red teaming and guardrail solutions under its flagship platforms, ARTEMIS and Repello Guard. The funding round includes participation from Venture Highway (now part of General Catalyst), pi Ventures, Entrepreneur First, and notable angel investors including Charles Songhurst, a Meta board member, Vivek Raghavan, CEO of Sarvam AI, and Satya Vyas, CEO of Project Hero. The company's core offering, ARTEMIS (Automated Red Teaming Engine for Mapping, Identification and Scanning), proactively scans GenAI models for vulnerabilities across modalities, including text, image, and audio. Complementing this is Repello Guard, a runtime security tool that filters unsafe outputs and flags risks like prompt injection, competitor mentions, and prompt leaks in real time. "We're at an inflection point where AI adoption is accelerating faster than security solutions can keep pace," said Aryaman Behera, co-founder and CEO of Repello. "Enterprises are deploying generative AI across every function, but they're doing it with yesterday's security playbook." In an exclusive interaction with AIM, Behera noted the rise of real-world failures, like a car dealership chatbot authorising $1 purchases, as evidence that AI safety cannot be an afterthought. He explained that it's like having a house without a door for generative AI applications without security. "You need such guardrails in order to protect your AI infrastructure," he added. Naman Mishra, co-founder and CTO, told AIM, "With ARTEMIS, we've turned red teaming, which used to be a quarterly enterprise task, into an integral part of the AI deployment pipeline when it comes to AI development." He highlighted that the goal of ARTEMIS and Repello Guard is to ensure fast and secure shipment of the products. Repello AI claims active deployments with firms like Groww and PhysicsWallah, and is currently running proof of concepts (POCs) across the US, Europe, and the Middle East. While specific revenue figures remain undisclosed, the new funding will be used to expand products, enhance brand credibility, scale go-to-market efforts, and form global partnerships.

AI security startup Repello AI raises $1.2 million from Venture Highway, Pi Ventures, others

Meta board member Charles Songhurst, Sarvam AI CEO Vivek Raghavan, and Satya Vas backed the company in this round. The fresh funds will be used to fast-track product development, expand the platform's red teaming and threat intelligence capabilities, and scale its go-to-market efforts.Artificial intelligence (AI)-backed security solutions startup Repello has raised $1.2 million in seed funding, with Venture Highway, Pi Ventures, and Entrepreneur First participating in the round. Angel investors, including Meta board member Charles Songhurst, Sarvam AI CEO Vivek Raghavan, and Satya Vas, have also backed the company in the round. Repello defends generative AI (GenAI) systems by using its products to simulate real-world attacks to find vulnerabilities and provide next steps for mitigation and safeguarding. Established in 2024 by IIT Roorkee alumni Aryaman Behera and Naman Mishra, Repello's engineering, product, and R&D teams are based in Bengaluru, while sales and marketing verticals are in San Francisco. The company plans to use the funding to accelerate product development and expand the platform's red teaming and threat intelligence capabilities and to scale its go-to-market efforts, Behera told ET. "We are also working on strategic partnerships and integrations to fuel the company's growth," he said. Behera affirmed the startup's belief in a proactive stance to cybersecurity, rather than reactive, with Mishra adding that their mindset has always been on offensive security to identify and address vulnerabilities in the evolving AI landscape. Repello AI's flagship product is ARTEMIS (Automated Red Teaming Engine for Mapping, Identification, and Scanning), which continuously runs millions of automated adversarial tests across text, image, and audio to deliver mitigation strategies to its customers. The other is Repello Guard, the company's threat monitoring and guardrails solution, which provides runtime security for AI applications across modalities, detecting AI-specific threats such as unsafe output, competitor mentions, and system prompt leaks. When asked about cyberthreats getting stronger due to AI, Behera said it is "a cat and mouse game." He said the defense side must learn from attacker techniques and use the attacker's perspective to develop defensive strategies. Repello's solutions are used by companies in India and some from the Fortune 500 as well. But North America is the largest market due to high AI adoption, but it has global customers in the Middle East as well.