AILA: Artificially Intelligent Laboratory Assistant

AILA (Artificially Intelligent Laboratory Assistant) is a streamlined framework designed for laboratory automation and assistance. This guide provides details on how to set up and run AILA on both Windows and macOS platforms, and where to find the benchmark data and results.

---

Table of Contents

1. Getting Started
2. Installation Instructions
   • Windows
   • Mac
3. Benchmark Data and Results
4. Support

---

Getting Started

AILA requires Python and the Streamlit library to run. Please ensure Python 3.10 or higher is installed on your system before proceeding. You can download it from python.org. It is highly recommended to use a new virtual environment.

---

Installation Instructions

Clone the Repository

git clone https://github.com/M3RG-IITD/AILA.git
cd AILA

Install Required Packages

Run the following command to install all necessary dependencies:

pip install streamlit matplotlib langchain_chroma langchain NSFopen scikit-image pymoo langchain_openai langgraph

Display AFM Bench Tasks

1. Navigate to the Data folder:

   cd Data

2. Run the AFM Bench Tasks using Streamlit:

   streamlit run task.py

   Wait a few minutes as it will take some time to load all the files.

Run AILA

1. Navigate back to the AILA main directory:

   cd ..

2. Enter the "app" folder:

   cd app

3. Run the AILA application:

   streamlit run app.py

4. LLM Model Setup:
   • Open the corresponding script in the "app" folder for the desired LLM model:
     • "AILA_claude-3-son.py" for Claude
     • "AILA_3.5.py" for GPT-3.5
     • "AILA_4.0.py" for GPT-4.0
     • "AILA_llama_3.3.py" for Llama
   • Search for YOUR_API_KEY in the script and paste your OpenAI API key into the embedding model and corresponding LLM model sections.

---

Benchmark Data and Results

The benchmark data and results for AILA are available in the Data and Results folders. Navigate to the following GitHub repository to access them:

• Data Folder: Contains all input data files used for benchmarking.
• Results Folder: Contains the output results of the AILA framework.

---

Citation

If you find our work useful, please cite

@article{mandal_evaluating_2025,
	title = {Evaluating large language model agents for automation of atomic force microscopy},
	volume = {16},
	issn = {2041-1723},
	url = {https://doi.org/10.1038/s41467-025-64105-7},
	doi = {10.1038/s41467-025-64105-7},
	number = {1},
	journal = {Nature Communications},
	author = {Mandal, Indrajeet and Soni, Jitendra and Zaki, Mohd and Smedskjaer, Morten M. and Wondraczek, Katrin and Wondraczek, Lothar and Gosvami, Nitya Nand and Krishnan, N. M. Anoop},
	month = oct,
	year = {2025},
	pages = {9104},
}

---

Support

If you encounter any issues or have questions, please open an issue on the GitHub Repository or contact the project maintainer.

---

Thank you for using AILA! We hope it enhances your laboratory workflows.