Realset AI, a real‑world data lab, and Flatkey announced a $10 million Series A to expand data capture for LLMs, frontier models and embodied agents.

Realset AI, a real‑world data lab, and Flatkey have secured a $10 million Series A round to accelerate the collection and curation of training data for large language models, frontier AI systems, and embodied agents.

Funding Details and Investors

The round was led by venture partners and strategic backers, with participation from existing investors who see growing demand for high‑quality, real‑world datasets that can bridge the gap between simulated training environments and physical deployment.

Purpose of the Investment

The capital will be used to expand Realset AI’s data capture infrastructure, enhance Flatkey’s platform for annotating and validating multimodal inputs, and build pipelines that feed directly into next‑generation AI models, including those powering autonomous robots and advanced conversational agents.

Both companies aim to address the scarcity of diverse, context‑rich data that frontier models require to improve reasoning, safety, and real‑world applicability.

Strategic Benefits for Frontier Models

By integrating Realset AI’s sensor‑driven data collection with Flatkey’s annotation tools, developers can obtain labeled datasets that reflect complex physical interactions, environmental variability, and human‑machine collaboration scenarios.

  • Enhanced training data for large language models (LLMs) that need real‑world grounding
  • Rich multimodal inputs for embodied agents such as drones, robots, and AR/VR systems
  • Improved safety testing through diverse, edge‑case scenarios captured in the field

Industry Context

The AI ecosystem is increasingly recognizing that synthetic data alone cannot fully prepare models for deployment in unpredictable environments. Real‑world datasets are becoming a critical differentiator for companies seeking to deliver reliable, trustworthy AI solutions.

Analysts note that investments in data‑centric AI startups have risen sharply, reflecting a broader shift toward building robust training pipelines that can support the next wave of AI capabilities.

High‑quality, real‑world data is the missing link for truly generalizable AI, and this funding will accelerate that journey.

The partnership between Realset AI and Flatkey positions them to become key suppliers of the data infrastructure needed for the emerging class of frontier models and embodied agents.

PR Newswire coverage of Realset AI and Flatkey Series A raise