Kuala lumpur: Sapient Intelligence has announced the launch of PRAXIST (Beta), an autonomous artificial intelligence (AI) research and development system designed to tackle complex technical challenges and independently test and validate potential solutions.

According to BERNAMA News Agency, William Chen, Co-Founder of Sapient Intelligence, stated that PRAXIST is engineered to provide organizations with additional research capacity. This allows them to explore technical problems with a breadth and speed that typically demands significantly greater specialist resources. The Singapore-headquartered artificial general intelligence research company emphasized that while AI has increasingly boosted productivity in content generation and other well-defined workflows, the domain of research and development (R and D) remains inherently resistant to automation and scaling.

The company explains that breakthroughs in R and D necessitate the testing of multiple hypotheses, learning from both successes and failures, and continuously assessing which paths warrant further pursuit. Organizations must navigate these challenges while balancing specialist expertise, time, cost, and infrastructure constraints. PRAXIST addresses these challenges by acting as an R and D capacity multiplier, autonomously exploring which technical approach can best achieve a measurable objective. Users are responsible for defining the goal, parameters, and budget, while PRAXIST conducts experiments and evaluates the most robust solution.

In a recent evaluation, PRAXIST achieved the highest-level result in 49 out of 75 challenging Kaggle competitions from the MLE-Bench, at a recorded model cost of approximately US$3,000. This performance is contrasted with Claude Code, which secured 34 highest-level results at an approximate cost of US$38,000 under the same conditions. MLE-Bench serves as a benchmark to assess how effectively AI systems address complex, real-world machine learning challenges.

PRAXIST employs multiple autonomous research peers to concurrently explore different approaches, test hypotheses, investigate failures, and validate promising results. It utilizes a generation-layered research graph that captures the insights from each experiment and builds stronger solutions based on discoveries made across various branches.

For organizations with limited AI or machine learning capabilities, PRAXIST can supply an AI research layer that complements existing domain expertise. For more advanced teams, it can enhance current capabilities, expanding the scale and breadth of R and D endeavors. Sapient Intelligence also noted that PRAXIST can function with proprietary data in private or customer-controlled environments, affording organizations greater control over their research infrastructure and sensitive intellectual property.