Ground AI-generated chemistry in quantum physics.

Chembricks checks the molecules AI models propose against quantum chemistry, before they reach the lab or feed further AI reasoning.

Open Chembricks AI Open to anyone. Ask about a molecule in the chat, or connect your own agents over MCP or the API.

AI proposes molecules faster than anyone can check them

Chembricks is for teams building AI agents, and for researchers in drug discovery, materials, metal recovery and water treatment. Each of its four layers can be used on its own.

Quantum reference data

High-accuracy quantum chemistry calculations (DFT and QMC), curated by quantum chemists.

Every other layer is built on it.

Free data subset on Hugging Face

Millisecond models

Our own models, trained on that data, predict energies and properties of organic molecules in milliseconds.

Each answer says how confident the model is, and whether the molecule is within what it was trained on.

Published benchmarks

Tools for AI agents

Our models plus 140+ chemistry tools, from pKa and solubility to spectra and retrosynthesis.

Any agent reaches them through one MCP connection, the standard way AI agents plug into tools, and every result shows its source.

Validated context

We feed checked results straight into an AI agent’s context while it reasons, each labelled with the method used and how certain it is.

The same results, with the steps behind them, are also available as training data for AI models.

Runs on ordinary CPUs

No GPU needed. Install the models privately in your own cloud, or reach them through agent platforms, marketplaces and data exchanges.

Calculations run for uncertain questions can be added to the reference data and, after testing, go into the next model version.