Realset AI and Flatkey Raised $10 Million
The companies secured Series A funding to expand training data operations for AI models and embodied agents.
Updated on Sept. 30, 2026 in Artificial Intelligence
Live Poll
Do you believe increased investment in AI training data will improve future technology safety?
Realset AI and Flatkey successfully closed a $10 million Series A funding round. This new capital will support the development of training data for frontier models and embodied agents.
Why it matters
The investment provides resources to grow a capture network of workplace environments and hire domain experts. These additions are designed to improve the quality of data used to train advanced artificial intelligence systems.
The $10 million in funding is earmarked for expanding specialized capture networks and increasing the workforce of expert demonstrators. This data is critical for training frontier models and embodied agents.
The players
Realset AI
Realset AI is a firm that produces specialized training data for large language models, frontier models, and embodied agents.
Flatkey
Flatkey is a partner firm that collaborated with Realset AI to secure the latest round of Series A funding.
The details
Realset AI specializes in producing training data tailored for large language models and embodied agents. The companies plan to utilize this funding to launch open benchmarks that measure AI policy performance.
Timeline
The Series A funding announcement occurred on September 30, 2026.
The Tech Race
This investment underscores the intense industry focus on sourcing high-quality, real-world data to advance the development of embodied agents. It marks a shift from general-purpose model training toward specialized data environments that facilitate complex AI interactions in physical spaces.
As this funding supports the creation of more accurate and capable embodied agents, users may experience more advanced automation in professional and studio settings. These developments are intended to improve how AI models interpret and interact with physical environments.
The takeaway
Reliable, high-fidelity training data remains a critical bottleneck for the next generation of AI development. Securing specialized capture environments will likely become a key differentiator for firms aiming to lead in model performance and safety benchmarks.
Further reading
For broader trends regarding the infrastructure behind generative systems, visit Artificial Intelligence.
Live Poll
Do you believe increased investment in AI training data will improve future technology safety?
