Toward autonomous chemical discovery: agentic AI in organic synthesis and nanochemistry

К автономному открытию химических соединений: агентный ИИ в органическом синтезе и нанохимии
Boppudi Hari Babu, Pullaiah Chowdary Vutla, Jagadeesh Mahimaluru, Bryce Ke Jun Chong, Srinivasulu Kambam
2026-09-24

agentic AIautonomous chemical discoveryautonomous laboratoriesnanochemistryorganic synthesis
Conventional approaches to chemical discovery rely heavily on human expertise and experimental trial and error. This makes the process slow and resource-intensive when exploring large chemical spaces and designing novel compounds. Artificial intelligence (AI) has transformed this field by enabling predictive, generative, and increasingly autonomous approaches. In particular, machine learning, deep learning, generative AI, and autonomous laboratories allow systems to predict chemical outcomes, design molecules and synthetic routes, and perform experiments in closed loops. Agentic AI extends these capabilities by allowing AI systems to plan multistep tasks, use external tools, interpret experimental results, and adapt subsequent actions to pursue scientific objectives. These capabilities are being integrated into autonomous AI systems across multiple fields, including organic synthesis, nanochemistry, computational chemistry, robotics, and autonomous experimentation. Together, these developments support a comprehensive review of agentic AI and its potential to enable autonomous chemical discovery. However, existing reviews have primarily examined organic synthesis and nanochemistry separately, without a unified framework for how agentic AI can integrate these domains toward autonomous discovery. In this review, we integrate advances in predictive and generative AI, agentic systems, autonomous laboratories, and related computational and robotic technologies, while distinguishing capabilities demonstrated experimentally from those proposed for future applications. We also examine the major challenges that must be addressed to achieve reliable autonomous discovery, including model interpretability, data quality, reproducibility, experimental validation, scalability, safety, and governance.
1
Agentic AI extends chemical AI by planning multistep tasks, using external tools, interpreting results, and adapting actions toward scientific objectives.
2
Predictive, generative, and agentic AI combined with autonomous laboratories can support closed-loop chemical discovery, including outcome prediction, molecule design, route planning, and experimentation.
3
Reliable autonomous discovery remains constrained by interpretability, data quality, reproducibility, experimental validation, scalability, safety, and governance challenges.
4
The review distinguishes experimentally demonstrated agentic capabilities from applications proposed for future development across chemistry, robotics, and autonomous experimentation.
5
The review unifies organic synthesis and nanochemistry within a framework for agentic AI-driven autonomous chemical discovery, addressing domains previously reviewed separately.

agentic AI systems for autonomous chemical discovery in organic synthesis and nanochemistry

integration of predictive and generative AI, multistep planning, tool use, experimental interpretation, and adaptive closed-loop experimentation, including challenges to reliable autonomous discovery

Publication Details
Publication Date
2026-09-24
Journal
Publisher
ISSN
Cited by
0
Access Type
Author Information
Authors
Boppudi Hari Babu
Pullaiah Chowdary Vutla
Jagadeesh Mahimaluru
Bryce Ke Jun Chong
Srinivasulu Kambam
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%