Sapient

Sapient Intelligence, a Singapore-headquartered AGI research company, is releasing PRAXIST (Beta), an autonomous AI research and development system that takes on complex technical problems and independently tests and validates potential solutions. Users set the objective, constraints, and budget, and the system then deploys multiple autonomous research peers to explore approaches in parallel, test hypotheses, investigate failures, and validate promising results . Initial applications are focused on sectors with intensive R&D requirements or highly measurable or simulatable optimization problems, including AI, engineering and robotics, manufacturing, finance, and health and drug discovery. The launch is aimed at a specific gap in the current agent landscape. Unlike general coding agents, which are built primarily to execute a specific task, PRAXIST is designed for long-horizon R&D, where the problem-solving approach itself may need to be discovered and adapted. Instead of the tree-like search used by similar systems, where each candidate inherits from one parent and weaker branches are pruned, PRAXIST uses a generation-layered research graph that preserves what every experiment teaches. This allows later generations to combine valuable mechanisms, evidence, and constraints across different lineages, including failed attempts. In internal evaluations, on 75 challenging Kaggle competitions from MLE-Bench, PRAXIST achieved the highest-level result in 49 competitions at an approximate recorded model cost of US$3,000, compared with 34 highest-level results for Claude Code at approximately US$38,000 under the same evaluation conditions. Partner tests are similarly concrete, with a rocket-landing simulation improved to 100% within 12 hours, and an industrial robotic SLAM problem where a partner team's accumulated error of 9.37cm after several months of development was reduced by PRAXIST to 5.01cm within three days. Sapient is led by founder and CEO Guan Wang , with co-founder William Chen and Jin Li as Chief Scientist of PRAXIST. The same team previously open-sourced the Hierarchical Reasoning Model (HRM), a brain-inspired architecture that leverages hierarchical structure and multi-timescale processing , and now operates with a global team of more than 50 researchers and engineers operating across Singapore, Palo Alto, and Beijing . PRAXIST can run inside customer-controlled environments, which matters for teams that want to point autonomous experimentation at proprietary data without handing it off.