Bench · Advanced · 25 min
Closing the Bench-to-Model Loop
BenchActive learning
Record lab results in the platform and let active learning pick your next experiments.
Steps
Follow along
Each step is something you can do in the platform right now.
01
Step 1
Create a recipe in the Bench Logger matching your standard synthesis protocol.
02
Step 2
Log your existing lab results against the recipe as you go.
03
Step 3
Run optimization on the logged data to build the search history.
04
Step 4
Open Active Learning and request the next batch of experiments.
05
Step 5
Verify which engine ran (Bayesian vs. heuristic fallback) in the response.
06
Step 6
Run the batch in the lab, log the results, and request the next round.
Notes
Pro tips
What experienced users wish they knew the first time.
Each round of real results sharpens the next proposal — the loop converges on high-information experiments.
Keep the recipe stable between rounds so the model sees consistent conditions.
Want the details? This tutorial pairs with the docs page stages · active learning.
Read the docs →