FormulaIQ
Inverse formulation
From NLP property briefs to recipes constrained by your inventory — with confidence and cost estimates.
- Property NLP to targets
- Inventory-aware component generation
- Custom models on your data
- Batch cost estimates
Platform
One platform, from molecule to plant.
FormulaIQ finds what to make. Studio Pro designs how to make it. Chemical Engineering Agents keep the plant on-spec — on a shared CAS-keyed chemical backbone.
From discovery to the plant floor.
One chemistry stack — not three disconnected tools.
Discover
FormulaIQ
Inverse formulation from target properties to in-spec recipes.
Design
Studio Pro
Process digital twins with validated thermodynamics.
Operate
Chemical Engineering Agents
Keep operations on-spec — and grow a fleet of chemistry AI agents.
The platform stack.
From natural-language property briefs to component generation, custom models on your data, full-metric process simulation, digital twin profiles, and control agents that take a simulated process into the plant — layered and inspectable, across industries.
Across industries
- Coatings
- Polymers
- Specialty chemicals
- Energy & batteries
- Adhesives
- Formulations
Simulate → operate
Same twin from simulation experiments to plant agents — full metrics in the simulator, digital twin profiles for handover, control agents on the floor.
- L5
Control agents
Simulate the process first, then hand the same twin to Chemical Engineering Agents that advise in the real plant — with human-in-the-loop for safety-critical decisions, and more agent roles ahead.
- L4
Process simulator & digital twin profiles
Build a full simulator with thermodynamics, kinetics, and live metrics. Run simulation experiments and extract digital twin profiles ready for scale-up and operations.
- L3
Formulation, properties & component generation
Inverse design from formula property targets — generate real components and in-spec recipes, not black-box suggestions.
- L2
Property NLP & prediction
Describe the product in natural language. NLP turns intent into property windows; physics-informed models predict with confidence intervals.
- L1
Data sources, models & workflow
Connect experimental and plant data sources, create or refine models on your chemistry, and orchestrate discovery → design → control in one workflow.
Same record · brief → plant
Watch the twin travel.
A living chemical record moves through the stack — NLP brief, formula, simulated twin, then plant agents — without losing CAS identity or uncertainty.
No translation layer. Uncertainty rides with the packet.
Chemical record
1 / 4
CAS 108-95-2viscosity94% conf
intent
Property brief
Natural language becomes a property window and market constraints.
Data sources & stock
Your inventory. Your recipes.
Connect your own inventory and material data sources. FormulaIQ designs formulations from what you already have on hand — then estimates cost so production stays grounded in real stock and real prices.
Optional — pilots work without inventory; connect when you want stock-aware design.
Inventory → FormulaIQ
step 1/4
Products
Explore each surface of the platform.
- Studio Pro
Process digital twins
Design and optimize processes with validated thermodynamics before capital is committed.
- Real thermo packages
- Full-metric process simulator
- Simulation experiments
- Digital twin profile extraction
- Chemical Engineering Agents
Plant on-spec — and a growing agent fleet
Hand a simulated process to chemistry-aware agents that keep the plant on-spec — with humans in the loop, and more agent roles on the same backbone ahead.
- Plant ops agents today
- Chemistry-aware advisory
- Human-in-the-loop
- More agents coming
Questions
- What is the Lavoisier platform?
- One chemistry stack connecting FormulaIQ (inverse formulation), Studio Pro (process digital twins), and Chemical Engineering Agents — sharing a CAS-keyed data backbone so discovery flows into process design and operations without translation.
- How is this different from black-box AI?
- Every prediction shows science baseline, data-driven correction, and quantified uncertainty. Chemists and engineers can challenge the answer — not just accept it.
- Do you use customer formulation data to train shared models?
- Customer formulation data is never used to train shared models without a written agreement. See our Security page for data handling details.
Bring a formulation or a process you care about.
We'll show you what the platform does with it — a 30-minute technical walkthrough on your chemistry, not a slide deck.