Self-serve onboarding workflow
From signup to your next recommended experiment.
PeakyPhi helps teams subscribe, upload experiments, define goals, and receive a practical recommendation without waiting for a sales-led setup.
Typical first session
Create account and choose a paid plan
Add historical data or 10–15 new runs and define the goal
Receive the next recommended experiment
How the optimization loop works
A practical loop for teams that need decisions, not statistical setup.
Experiments update the model. The model recommends the next run. Your team tests it, uploads the result, and the loop gets smarter.
Experiments
→
Predictive model
→
Recommendation
→
Run experiment
↺
Upload result
01 · Create a project
Define the process.
Name the project, choose the process area, and list the controllable settings your team can change: temperature, pressure, residence time, catalyst loading, pH, feed ratio, machine speed, or any other input.
Typical inputs: controllable settings, allowed ranges, units, known constraints, and optional notes for operators.
02 · Add known experiment results
Upload your first experiments.
Enter or import the experiments you already ran. Each row connects input settings to measured outcomes so the model can estimate which areas of the process space are promising and which are uncertain.
Useful outputs: yield, purity, defect rate, cycle time, energy use, material cost, viscosity, strength, or throughput.
03 · Set optimization goals
Set the goal.
Choose one or multiple objectives. Maximize yield while minimizing cost, keep impurity below a threshold, or search for a process window that balances quality and throughput.
Goal types: maximize, minimize, hit a target, stay within range, or respect hard constraints.
04 · Get the next suggested experiment
Review the recommended run.
The model proposes the process settings most likely to improve your goals or reduce uncertainty. Your team sees the suggested values, expected tradeoffs, confidence, and why this experiment is useful.
Recommended Experiment
Temperature 184°C
pH 6.8
Time 42 min
Predicted Yield 92.4%
Confidence High
Expected Improvement +7.8%
05 · Run, upload, repeat
Run, upload, repeat.
Run the suggested experiment, add the result, and the model immediately updates. Each cycle makes future recommendations more targeted and reduces wasted trials.
The loop continues until you find a strong operating window, reach a target, or decide the remaining improvement is not worth another run.
What users enter
The app turns domain knowledge into a usable optimization model.
Users do not need to describe the mathematics. They describe the experiment: what can be changed, what should be measured, and what tradeoffs matter.
Process variables
Inputs the user can control: temperature, time, pressure, pH, material ratio, machine speed, or flow rate.
Measured outcomes
Results from each experiment: yield, purity, cost, cycle time, defect rate, viscosity, strength, or throughput.
Constraints and priorities
Ranges the model must respect and goals it should balance, including hard limits and preferred operating regions.
Ready for the first run
Create an account, subscribe, and start building your optimization loop.
Once payment is active, users can create a project, enter experiment data, define goals, and receive the next suggested experiment directly inside the app.