Thoth AI published a continual learning result showing that when a policy learns a new task, replaying a buffer of past experience can cause catastrophic forgetting on high-conflict tasks, dropping retention by 63% when memory anchors are included.

Thoth AI is a global AI data solutions company with R&D operations in Silicon Valley, working with AI teams to support the design, development, and deployment of production-grade AI systems. We help our partners build and scale high-quality datasets and evaluation frameworks for advanced models, including LLMs, VLMs, multimodal systems, and applied AI used in real-world products, such as robotics. Our focus is on the areas that most directly impact model outcomes: data quality, edge-case coverage, and robust evaluation. Through a combination of technical expertise, disciplined processes, and global delivery capabilities, we help organizations deploy AI systems that are reliable, scalable, and ready for real-world use. Mission To help organizations build and deploy reliable AI systems by providing high-quality data, evaluation, and operational support that meet production requirements. Vision To be a trusted global partner for AI teams, recognized for enabling safe, reliable, and scalable AI deployments across industries and applications. Values Quality & Reliability We prioritize accuracy, consistency, and robustness in every dataset and evaluation workflow. Responsible Execution We operate with strong governance, clear processes, and respect for client, user, and regulatory requirements. Collaboration We work closely with our partners to align on goals, constraints, and timelines. Practical Innovation We apply research and technical advances in ways that deliver measurable, real-world value. Global Delivery Excellence We combine technical leadership with scalable, dependable global operations.