Press release

AutoAlign launches first generative AI Sidecar to bolster AI safety

AutoAlign spins out of Armilla AI with Sidecar, a dynamic LLM security firewall, and a cooperation agreement with KPMG in Canada.

Stylised illustration of a motorcycle whose sidecar carries a lime shield

Toronto, April 23, 2024. AutoAlign, an AI safety company providing robust generative AI security, announced its spin-off from Armilla AI, the launch of Sidecar (its dynamic LLM security firewall solution), and a cooperation agreement with KPMG in Canada to strengthen AI-model risk frameworks. The foundational Sidecar technology operates alongside LLMs, delivering consistent security through dynamic interaction while evolving with rapidly advancing language models. KPMG will incorporate AutoAlign as a compliance option within its Trusted AI framework, enabling businesses across sectors to configure Sidecar for their compliance and security requirements when deploying trustworthy generative AI.

By the time companies build AI with proper security guardrails, they’re immediately obsolete. That’s why we built Sidecar — so enterprises have LLM security that is comprehensive, nimble, and always evolving.
Dan AdamsonInterim CEO and co-founder, AutoAlign

Adamson explained the separation: when generative AI exploded, it became clear that Armilla’s two business anchors (model security and AI insurance) required their own companies to provide deserved customer services.

Generative AI’s rapid expansion has created challenges including hallucinations, jailbreaks, and bias. AutoAlign developed Sidecar as the first firewall security solution that directly interfaces with LLMs to address these issues. The system’s adjacent rail structure with contextual Alignment Controls operates parallel to models, seamlessly transitioning across models and use cases. When Sidecar rejects a response, it prompts the model to make corrections, improving model accuracy through real-time learning while refining outcomes for specific business applications. This approach allows enterprises to prioritise performance enhancement rather than over-tuning model capabilities.

Working with KPMG exemplifies two enterprises committed to AI’s powerful and safe future.
Rahm HafizCTO and co-founder, AutoAlign
AutoAlign shares our goal of integrating generative AI into businesses in a responsible manner without compromising the effectiveness of this transformative technology.
Ven AdamovPartner, National Leader of Risk Analytics and Trusted AI co-leader, KPMG in Canada

Get started

Ready to talk?

If operational intelligence matters to your production line, your field service operation, or your programme, we want to talk.