From instrument output to the next experiment
Discovery Loop removes the researcher from the loop between runs. Each experiment feeds the system. Each cycle creates a prioritized next step. Here is how it works.
Link your instruments
Discovery Loop includes adapter modules for common laboratory instrument families, including HPLC, plate readers, spectroscopy systems, and sequencers. Set each adapter to your instrument's file output path or API credentials. After that, every completed run enters the loop automatically.
For unsupported formats, your team can use the open adapter SDK to create a parser. Each adapter writes to one normalized schema, so adding an instrument type does not require changes elsewhere in the loop.
File watch or API
An adapter can monitor an output folder for new files or poll an instrument API at an interval you set.
Lab or cloud
Run adapters on a workstation beside the instruments, or use a cloud instance when your data already leaves the lab.
Set the research objective
Start by telling Discovery Loop what you want to find. You express this as an objective function: a value to maximize or minimize, a target compound yield, a stability threshold, or a structural property. The system is not tied to one field. Any numeric result measured by your instruments can serve as the objective.
Define the parameter space: the variables the loop can manipulate (temperature, pH, concentration, catalyst choice, reaction time) and their allowed ranges. The loop searches there.
Interpret every run
Once a run finishes, its adapter sends parsed results to the analysis module. That module identifies the signal relevant to your objective function, assigns a confidence interval, and updates the hypothesis tree. This branching structure records which parts of the parameter space are understood and which remain uncertain.
The hypothesis tree carries forward from one cycle to the next. As evidence builds, the loop develops a clearer view of the search space.
Search space explored: 62 of 180 candidates
Choose the next experiment
The decision engine evaluates every candidate still open in the hypothesis space. Its scoring policy seeks the greatest expected information gain for the estimated run cost. That reflects a deliberate scientific priority: learn as much as possible for each dollar of wetlab spend, rather than minimize expected regret in the RL sense.
The highest scoring candidate becomes a protocol in your lab's experiment template, then moves to the instrument queue. No human approval or scheduling is required for the next step. You can still configure an approval gate when a researcher should remain in the loop for particular instrument types.
Inspect cycle history anytime
The loop records every decision and its reasoning: the evidence behind it, the candidates left aside, and the information gain scores at that moment. Replay any cycle to understand why one experiment was chosen over another.
The cycle log is also the output artifact of a completed research project. When the loop reaches an optimum or exhausts the hypothesis space, the log records the full path from first run to conclusion.
See Discovery Loop in your research setting
Talk with the team. We will show how the loop fits your instruments and objective function.