Deccan AI, an AI data and evaluation startup, has raised $25 million in a funding round led by A91 Partners. The round also saw participation from existing and global investors, including Susquehanna International Group (SIG) and Prosus Ventures. The company focuses on building systems that improve AI model training through high-quality data generation, human feedback, and evaluation workflows. It works to improve the accuracy and reliability of large language models and enterprise AI tools.
Unlike companies that focus on building end-user AI applications and chatbots, Deccan AI operates in a more foundational segment of the industry, focusing on building systems that improve how AI models are trained, evaluated, and refined using structured human intelligence. This segment has become significant as large language models are now widely deployed across industries like finance and software development, which require higher accuracy and domain-specific reliability.
“Everyone is busy hyping ‘Super Intelligent AI’, but we are obsessed with something far more important in our view: ‘Super Accurate AI’. Accuracy is the cornerstone of truly usable AI and drives our work across data/environments/agents,” Rukesh Reddy (Founder, Deccan AI) noted.
The latest funding will primarily be deployed to expand the company’s enterprise AI capabilities and strengthen its core platform that allows high-precision data generation and model evaluation workflows. This includes scaling operations that rely on expert human contributors who help improve the accuracy of AI outputs, validate model responses, and support learning processes where human feedback is essential. The company’s model is based on combining distributed human expertise with AI tooling to create more reliable datasets and improve model performance in real-world applications.
“This round will help us invest further into some of the most important problems in the frontier model space today, e.g. RL environments, complex expert trajectories. Additionally, we are excited to double down on Enterprise and Physical Intelligence, areas where we are seeing tremendous early traction,” Rukesh added.
Founded in October 2024, the company claims that its customer base already includes prominent AI and cloud industry players like Google DeepMind and Snowflake. The company operates through a large distributed contributor network that is central to its model of ‘human-in-the-loop’ AI development. According to the firm, it currently has over 1 million contributors, ranging from students to domain experts, who participate in tasks like data labelling, model evaluation, and structured feedback generation.
The timing of the funding is particularly significant as the global market for AI training services and data labelling is expanding rapidly, along with the rise of generative AI and LLMs. The global data labelling market alone is projected to surpass $6 billion by 2030. The AI training dataset market is also scaling quickly, with projections showing it could grow to over $16 billion by 2030.
The Tech Portal is published by Blue Box Media Private Limited. Our investors have no influence over our reporting. Read our full Ownership and Funding Disclosure →
