Scaled Cognition, an AI lab, recently raised a $100M Series A to develop their APT (Agentic Pretrained Transformer) model, which is designed to deliver human-like conversational fluency while eliminating hallucinations for high stakes enterprise actions.
Scaled Cognition announced a $100 million Series A funding round, led by Khosla Ventures with participation from Genesys. The round values the company at approximately $750–850 million post money, reflecting strong investor conviction in its specialized approach to enterprise AI despite the competitive landscape.
What is Scaled Cognition?
Scaled Cognition, based in Mountain View, California, develops Super-Reliable Intelligence for customer experience (CX) and high stakes enterprise workflows. Its flagship model, APT (Agentic Pretrained Transformer), targets the core limitation of general purpose LLMs: hallucinations and unreliability in production environments.
APT delivers conversational fluency comparable to leading frontier models but guarantees:
- Zero hallucinations through architectural innovations rather than post-hoc guardrails.
- Policy fidelity: enforced permissions, constraints, and business logic that cannot be broken.
- Verifiable action execution via structured predictions (e.g., programs and system queries) alongside token streams.
- Model agnostic, open design: stateless RESTful API, self hosted/VPC/on-premises deployment, with prompts and logic staying under customer control.
The company provides a full platform including agent authoring tools (no code/low code/SDK), GenAPI™ simulation for testing, agent evaluators, and production monitoring to detect drift and violations. This enables rapid POC to production timelines (days rather than months) with provable reliability before deployment.
Target markets focus on CX in banking/finance, retail/e-commerce, travel/hospitality, telecom, insurance, healthcare, and BPOs. It emphasizes high containment, CSAT improvement, compliance, and cost savings by resolving more issues autonomously. The broader vision extends to insourcing parts of the $600 billion BPO market with owned AI workforces.
Genesys already integrates APT into its Genesys Cloud platform for agentic virtual agents, and the company claims production use with Fortune 500 customers. It projects over one billion customer service interactions automated in the next 12 months via its models.
Who are Scaled Cognition’s leaders?
- CEO Dan Roth: Serial entrepreneur with prior experience as Microsoft Corporate VP of Conversational AI. Co-founded Semantic Machines (pioneering agentic AI), acquired by Microsoft in 2018.
- CTO Dan Klein: UC Berkeley professor and leading NLP researcher.
- CFO Damon Pender: Also involved in the prior Semantic Machines venture.
The team’s deep expertise in agentic systems and enterprise deployment gives them credibility in addressing reliability from first principles. They argue that current LLMs are “plausibility engines” optimized for fluent output rather than truth, necessitating a new architecture for narrow domain, high stakes applications.
This Series A follows earlier seed/early funding (including a reported ~$21M round involving Lerer Hippeau, SilverCircle, and others). The $100M infusion is substantial for a specialized AI lab at this stage and will primarily:
- Expand the research team (PhD researchers and engineers).
- Accelerate enterprise deployments and commercialization.
Vinod Khosla highlighted the research heavy approach and developmental risk taken to build reliability into the core architecture, contrasting it with layering solutions on frontier models. The strategic investment from Genesys signals validation in the CX space.
Recommended: Chptr Raises $5.5 Million In Series A Funding Round
In a market saturated with general LLMs and agent frameworks, Scaled Cognition differentiates through deterministic reliability engineered from the ground up. Key advantages include:
- Suitability for regulated, action oriented domains where errors have real costs (financial transactions, medical records, claims, etc.).
- Reduced vendor lock-in and deployment flexibility.
- Comprehensive tooling for simulation, evaluation, and monitoring, lowering enterprise risk and speeding time to value.
Challenges in the space remain significant: scaling reliable agents beyond CX, competing with hyperscalers and well funded generalists, and proving consistent ROI at massive scale. However, the focus on a “reliability gap” in enterprise AI aligns with growing customer frustration over hallucinations in production.
The round underscores continued investor appetite for differentiated AI infrastructure plays that solve specific, painful enterprise problems rather than chasing general intelligence. At a ~$750M+ valuation post Series A, Scaled Cognition positions itself as a high potential specialist in the agentic AI wave, leveraging proven founders, strategic partnerships (e.g., Genesys), and early production traction. Funds will likely fuel talent acquisition in a competitive AI hiring market and further R&D to expand beyond CX.
Overall, this funding validates the thesis that reliability, not just capability, will be a decisive factor in enterprise AI adoption, particularly for customer facing and operational automation. The company is well placed to capture value in high stakes CX and BPO transformation.
Please email us your feedback and news tips at hello(at)superbcrew.com
Activate Social Media:
