ENGINEERING EXPLAINED
One good run is a starting point.
Keep the runs, compare their conditions, and make the next evidence gap visible before choosing pilot equipment.

By Nathan, chemical engineer and co-founder of RARE Labs.
KEEP THE RAW RUNS
Build an evidence map.
Start with the invented example or clear it for your own review. Feed labels identify comparable material and concentration. Temperature is °C, time is minutes, and outcome is recovered yield on feed in percent. Change the outcome basis in your downloaded notes if your process needs a different measure.
| Run / note | Feed condition | Temp / °C | Time / min | Yield / % | Observed result |
|---|
Table data stays in your browser. Pass/fail is your recorded result, not an automatic equipment recommendation.
Within-condition observations
Points show tested conditions. Connecting them would not establish that every condition in between works. Repeats at the same coordinates appear as concentric rings.
One good run starts the review.
Record the feed condition, temperature, residence time and outcome on one consistent basis. Keep failed runs and distinguish controlled variables, measurements and assumptions. The relevant variables depend on the chemistry and equipment.
Statistics shown are descriptive. A single observation cannot estimate repeatability. Sample standard deviation uses n − 1 and is reported only within an intentionally comparable condition group. The observed pass count is not a reliability guarantee; neither a small spread nor a passing run validates scale-up.
Primary sources: NIST — experimental-design principles · NIST — planning experiments · NIST — measures of scale.
RARE LABS
From lab chemistry to production.
We are building an autonomous scale-up system for chemistry using robotics and AI, starting with peptides and pharma.
