AI model lab Scaled Cognition has successfully raised $100 million in a Series A funding round to address the critical issue of reliability in artificial intelligence. The round was led by Khosla Ventures and aims to accelerate the company's mission of building AI systems that enterprises can trust for critical operations. This new capital will be used to expand the company's research team and expedite the deployment of its technology across major industries.
Tackling the Enterprise AI Reliability Gap
While modern AI has demonstrated impressive capabilities, its adoption in high-stakes enterprise environments remains limited by a fundamental lack of reliability. Issues such as hallucinations and factual errors prevent AI from being used in workflows where mistakes have significant consequences. Scaled Cognition was founded to solve this challenge, targeting sectors like financial services, healthcare, and insurance where accuracy is paramount.
The company was co-founded by CEO Dan Roth and CTO Dan Klein, a distinguished AI professor at UC Berkeley, who previously built and sold an agentic AI company to Microsoft. According to Roth, the problem is not a matter of resources but of architecture, requiring years of methodical work to overcome. Their focus is on creating systems that perform consistently and correctly, moving beyond the spectacular but often flawed demonstrations of current models.
A New Architectural Approach to Trustworthy AI
Scaled Cognition's flagship model, the Agentic Pretrained Transformer (APT), is designed to deliver the conversational quality of leading models while completely eliminating hallucinations. This concept, which the company calls "Super-Reliable Intelligence," ensures that the AI adheres strictly to predefined policies. The model is also engineered to be smaller, faster, and more cost-effective than frontier models, offering enterprises a more efficient solution.
A key differentiator is that reliability is engineered directly into the model's architecture rather than being an add-on feature. Klein notes that the most dangerous errors are those that appear correct, a problem their technology is built to solve. This approach allows enterprises to deploy the model in their own virtual private cloud or on-premise, giving them full ownership and control over their AI infrastructure.
Strategic Backing and Market Application
The investment from Khosla Ventures underscores the industry's need for a new approach to AI reliability. Vinod Khosla, the firm's founding partner, praised Scaled Cognition for taking on the developmental risk of creating a fundamentally different architecture instead of simply layering on top of existing models. This research-intensive strategy has resulted in a model that is designed to avoid providing incorrect answers in critical situations.
The technology is already seeing significant market validation through a partnership with Genesys, a global leader in AI-powered customer experience. Genesys utilizes the APT model within its Genesys Cloud™ platform for its virtual agent capabilities and has also invested in Scaled Cognition. This collaboration highlights the immediate applicability of Super-Reliable Intelligence in automating customer service operations for thousands of organizations worldwide.
The Future of Enterprise Automation
Beyond the initial application in customer experience, Scaled Cognition is targeting the expansive $600 billion business process outsourcing market. The company's vision is to empower enterprises to insource critical functions like IT support, HR, and finance by building their own AI workforces. This shift would give companies greater control and efficiency over processes that were previously managed by third-party services.
To facilitate this, Scaled Cognition offers a comprehensive platform that includes agentic tooling, simulation and evaluation frameworks, and live agent monitoring. This full-stack solution provides enterprises with everything needed to build, test, and operate reliable AI at a significant scale. Companies using the platform are on track to automate over one billion customer service interactions within the next twelve months, driving substantial operational savings.
In conclusion, Scaled Cognition's $100 million funding marks a significant step toward making AI a truly dependable tool for the enterprise. By focusing on an architectural solution to the problem of reliability, the company is poised to unlock new opportunities for automation in critical business functions. Its Super-Reliable Intelligence promises to provide the trust and consistency necessary for AI to handle tasks where the stakes are genuinely high.