Researchers at IIT Delhi, in collaboration with teams from Denmark and Germany, have developed AILA (Artificially Intelligent Lab Assistant) an AI agent capable of independently operating complex laboratory equipment, designing experiments, making real-time decisions, and analyzing results. Demonstrated on an Atomic Force Microscope (AFM), AILA marks a paradigm shift from AI-assisted to fully autonomous scientific research.
Glimpse:
Published in Nature Communications (December 2025), AILA automates nanoscale materials experiments using AFM, reducing tasks that once took a full day to just 7-10 minutes. Unlike traditional AI tools for writing or data analysis, AILA physically interacts with lab instruments, adapts to conditions, and interprets outcomes accelerating discovery in materials science while highlighting needs for safety safeguards.
Researchers at the Indian Institute of Technology (IIT) Delhi have achieved a groundbreaking milestone in laboratory automation with the development of AILA an AI agent that can autonomously conduct real-world scientific experiments. This innovation pushes AI beyond digital assistance into physical laboratories, performing tasks akin to a trained human scientist.
At the core of the demonstration is the Atomic Force Microscope (AFM), a highly sensitive instrument used to study materials at the nanoscale. Mastering AFM typically requires years of expertise in physics, surface interactions, and real-time control. AILA independently controls the device, optimizes parameters, responds to live feedback, collects data, and analyzes resultsย all without human intervention.
Key performance gains include slashing optimization time for high-resolution imaging from a full day to 7-10 minutes, enabling faster research cycles. The system uses an agentic AI framework with large language models to translate natural language instructions into executable actions, adapting dynamically to experimental variations.
The international collaboration involved IIT Delhi researchers (including first author Indrajeet Mandal, Jitendra Soni, and Zaki), Aalborg University (Denmark), Leibniz Institute of Photonic Technology (Germany), and University of Jena (Germany). Supervisors Prof. Nitya Nand Gosvami and Prof. NM Anoop Krishnan emphasized the shift: AI now “does science” rather than just supporting it.
Challenges noted include bridging theoretical AI proficiency with real-lab adaptability and implementing safeguards against deviations that could damage equipment. Future plans involve scaling to multiple instruments and developing indigenous models to reduce reliance on commercial LLMs.
This breakthrough aligns with India’s AI for Science initiatives, promising to democratize advanced research, boost productivity in fields like energy storage and sustainable materials, and position Indian institutions at the forefront of autonomous discovery
โAILA helps me with my daily experimental tasks and speeds up my research progress significantly. Previously, it would take a full day now the same task is completed in just 7โ10 minutes.โ
By
HB Team
