Skip to content

A data-driven platform for materials R&D

Connect experiment, simulation and literature data for more efficient experiments and smarter decisions.

I. Overview

Scattered data

Literature, experiment and simulation data are scattered across files, spreadsheets and software, making relationships difficult to preserve.

Hard to reproduce

Incomplete formulations, process parameters, raw data and processing records make results difficult to trace.

Knowledge is hard to retain

Research experience, methods and literature evidence are difficult to find and share, so teams repeat the same work.

Room to improve R&D efficiency

Manual searches, repeated processing and trial and error make it difficult to use existing data for the next design cycle.

Goal

Build a data-driven closed-loop platform for intelligent materials design, with more efficient experiments and smarter decisions.

II. Features

1. Experiment data collection

Start with materials and formulations, record preparation and testing, build a domain database, standardize processing and create Origin plots in batches.

Every material, with a clear source.

Maintain materials, properties and sources together, from chemicals to fibers and foams. Formulations and sample preparation reuse the same material records.

  • PubChem lookup and custom entry
  • Properties, suppliers, grades and batches
  • Reuse in formulations and sample compositions
Read operating instructionsGo to: User manual › Build a material library
Build a material libraryExample data
Workflow demo · Add material
MATERIAL / M–024

Matrix resin

Editing
Name / gradeMatrix resin · example grade
Solids contentNeeds review
Batch / sourceTo be added
Keep properties, batch and source in one record

Interactive workflow with example data. Actual operations depend on your installed version.

2. Simulation data collection

In development

Link simulation results to research objects, calculation conditions and experimental evidence.

Planned records include material structures, methods, software versions, boundary conditions, inputs and outputs. Supported formats and operating pages are still in development.

  • Calculation conditions and input files
  • Simulation results, units and sources
  • Links to materials, samples or research tasks
Read the user manualGo to: User manual › Simulation data collection

3. Literature data collection

In development

Turn evidence in publications into searchable domain data.

Collect publication metadata and sources; identify materials, formulations, processes and properties; then create structured records after human review. Extraction fields and import workflows are still in development.

  • Title, authors, year and DOI
  • Materials, experiment conditions and properties
  • Original locations and human review results
Read the user manualGo to: User manual › Literature data collection

4. Data center

Under construction

Bring scattered sources into a traceable domain database with controlled access.

Unify fields, relationships, revisions, permissions and search across experiment, simulation and literature data. The unified data center is under construction; installed laboratory services depend on actual configuration.

  • Consistent data structures and source relationships
  • Versions, permissions and controlled search
  • A data foundation for knowledge assistants and materials models
Read the user manualGo to: User manual › Data center

5. Large Language Model (LLM) + RAG knowledge assistant

Planned

Help people use existing knowledge more efficiently.

Use accumulated domain data with large language models and retrieval-augmented generation (RAG) to find experiment records and similar proposals, then produce reports with verifiable data and publication sources.

  • Search authorized records by research question
  • Compare similar conditions, results and literature evidence
  • Generate experiment proposal reports for researchers to review
Read the user manualGo to: User manual › LLM + RAG knowledge assistant

6. Machine Learning (ML) materials agent

Planned

Learn patterns from data for property prediction and inverse design.

Develop and validate domain models using accumulated data for property prediction, virtual screening, multi-objective optimization and experiment design. Iterate through model recommendations, experimental validation and data feedback.

  • Property prediction and target-driven inverse design
  • Virtual screening and multi-objective optimization
  • Candidate experiments, validation and feedback
Read the user manualGo to: User manual › ML materials agent

III. Questions & answers

Can I use only experiment data collection and processing?

Start with materials, formulations or instrument files, then add sample and source records as needed. Available modules and plugins depend on your installed version and authorization.

Are simulation, literature, LLM and ML features available now?

Simulation and literature collection are in development, the data center is under construction, and the LLM/RAG and ML agents are planned. These pages describe goals and expected inputs and outputs; release notes define availability.

How do LLM + RAG and the ML materials agent differ?

LLM + RAG helps researchers find, understand and reuse existing knowledge and prepare experiment proposals. ML learns patterns in structured data for prediction, screening and inverse design. They share a data foundation but perform different tasks.

Does plotting in Origin automatically update experiment records?

Versions with result archiving can link Origin projects, images and participating samples. External changes require a manual result-version update. Direct Origin transfer requires a compatible local environment.

Can team templates and experiment data have separate permissions?

Access can be managed by resource and member. Sharing methods and sharing raw research data require separate authorization. Uploading does not make a resource public, and feature licensing does not grant data access.

Where can I find detailed instructions and videos?

The manual covers six feature groups, including seven experiment chapters. Each chapter links to its matching Help center resource. Videos will be added after recording.

Go to the Help center

Built for experiment data workflows.