The problem: repeating tests you've already run
Most formulation teams repeat tests they've already run. Past results sit in spreadsheets, PDFs and notebooks, so nobody can see what was tried before, and the next experiment ends up being a guess.
What this demo shows
An all-purpose cleaner, APC-112, misses its cost target at $0.58/kg. Its ingredients, lots and test results sit in one record, and a PDF test report is read straight into the data. The team sets targets of 200–600 cP viscosity, at least 85% cleaning score and no more than $0.50/kg.
- One record per trial. Ingredients, lots, process settings and results sit together, so every trial adds to what the model knows.
- Five recommended formulations. Each comes with predicted viscosity, cleaning score and cost, all aimed at the targets.
- What drives the results. Effect sizes show that SLES 70% moves viscosity the most.
- Bodie, Uncountable's AI agent, digs in. Asked for past trials, Bodie charts 25 earlier SLES 70% runs and returns five options predicted to come in under $0.50/kg.
- Scientists stay in control. The team reviews the options and decides what actually gets tested.
Why it matters
The result is fewer repeated tests, faster decisions, and lab learning that isn't lost when people move on.
Everything happens in one connected record instead of a stack of separate tools, so the model, the effect sizes and Bodie all work from the same trials your team already ran.
See it with your own data
This video is part of DOE and predictive tools. To see how it would work with your own formulations and records, request a demo. Our team replies within 30 minutes during business hours.
