Design of Experiments, made approachable
Plan smarter experiments. Find the optimum faster.
Describe what you want to learn. Covariance Labs plans the experiment, then analyzes your results with real statistics: ANOVA, lack of fit, diagnostics and response surfaces.
Free to start, no credit card. Sign up with Google or an email link. Or try the free design generator first, no account needed.
From question to optimum in three steps
Describe
Say what you are trying to improve in plain English: “Maximize yield by varying temperature 60–90 °C and pressure 1–5 bar.” You review and edit the factors before anything is built.
Run
Get a randomized run sheet using the right design for your factor count. Do the experiments and type or paste the results in, straight from a spreadsheet if you like.
Understand
A fitted model, ANOVA, diagnostics and plain-English findings update as each result goes in. Then find the settings that hit your target.
Everything you need to plan and analyze an experiment
Claude-assisted design
Turn a sentence into factors, ranges and a goal. Every field is editable, and if the AI service is unavailable a built-in parser takes over.
The standard designs
Full and fractional factorials, Box-Behnken and central composite designs for up to six factors, with 1–10 centre points.
Real statistics
Least-squares fit, ANOVA with lack of fit, R², adjusted and predicted R², and studentized-residual diagnostics, all computed in the app.
Optimum finder
Maximize, minimize or hit a target value, and see where in your factor space the fitted model says to run next. Pro.
Spreadsheet friendly
Paste results from Excel, export the run sheet as CSV, and keep full backups as JSON.
Saved and in sync
Sign in with an email link to keep experiments in your account. Edits from two devices are detected, never silently overwritten.
Numbers you can check
The statistics are written from scratch and tested: probability functions are compared against published critical values, the regression against a standard benchmark dataset, and designs against their defining properties. The AI never touches the maths.
Read the methodsBuilt for people who run experiments in
- Chemical and process development
- Formulation and materials
- Bioprocess and cell culture
- Food and agriculture
- Manufacturing and quality engineering
- R&D
- Academic research
Free to design and analyze
Planning, the run sheet and the full statistical analysis cost nothing. Pro adds the 3D response surface and optimum finder.
Questions
Do I need to know statistics to use it?
No. You describe the goal and review a plan, and results are explained in plain language. If you do know the theory, the full ANOVA table, coefficients and diagnostics are all there, and the Methods page says exactly how each number is computed.
What is Design of Experiments?
It is a way of changing several factors at once, in a planned pattern, so that a small number of runs reveals each factor's effect and the interactions between them. That usually takes far fewer runs than changing one factor at a time.
Which designs are supported?
Full factorial, fractional factorial, Box-Behnken and central composite designs, for two to six factors. The app suggests one based on how many factors you have and what you want to learn, and you can change it.
Does the AI do the statistics?
No. The AI only reads your description and proposes factors, ranges and a goal. The design and all analysis are deterministic code you can inspect on the Methods page, and every proposal is checked and editable before use.
Where is my data stored?
In your account, in our database, protected so that only you can read them. You can export everything as CSV or JSON at any time. Descriptions you type into the AI step are sent to Anthropic to be parsed. See the Privacy page for details.
Is it free?
Designing experiments, the run sheet, and the full analysis are free. The 3D response surface and optimum finder are part of Pro, and paid billing is not open yet.