We optimize chemical processes
with AI
Increase yield, margin, and energy efficiency with recommendations grounded in the simulators, constraints, and plant data your engineers already trust.
What plants are up against
Chemical plants face shifting conditions, competing performance targets, and growing pressure to operate sustainably.
Plants never operate in a single steady state
Feedstock composition varies over time, ambient conditions vary with weather and seasons, and upstream disturbances cascade through the process. Whether the response requires updated steady-state setpoints or a dynamic trajectory adjustment, operations must adapt continuously. One-off optimization studies capture a snapshot, but the plant has already moved on.
- Feedstock variability in composition, flow rate, and quality
- Upstream and downstream upsets that propagate through units
- Conditions shift faster than periodic optimization studies can keep up
Your simulator. Our AI. Better operations.
We build digital twins from the tools your engineers already trust, then use AI to find operating improvements that manual studies can't reach.
We build a validated digital twin from your simulator
Starting from CHEMCAD, Aspen, or the simulator your team already uses, we create a calibrated digital twin of your process. Your operating data shapes the model. No new software, no black boxes.
- Works with your existing process simulator
- Calibrated against your historical plant data
- Your team keeps the twin after the project
From data to optimized operations
We start from your existing simulator and operating data, define the objective with your team, and return recommendations your engineers can review and validate.
Build the twin
- Build and validate a digital twin using the simulators and data your team already trusts.
- Define the objective, constraints, and safety limits with your team.
- Set the decision variables and operating ranges the model should explore.
Train and test
- Train the AI agent across simulated operating conditions until the policy converges.
- Test performance across disturbances, constraints, and changing feed conditions.
Deliver insights
- Deliver controller setpoint recommendations and improved operating conditions.
- Package the digital twin and insights so your team can reuse them after the project.
Built on tools you trust. Every recommendation is grounded in your real process model, so engineers can review, validate, and act on the results with confidence.
Real results
A practical operating view that ties optimization behavior back to the units, streams, and economics your team already monitors.
Best Reward
Optimization outcome
Engineers can see whether the policy is still learning or has stabilized around better setpoints.
Recommendations your engineers can review and validate
Not a black-box score. You get setpoint recommendations, expected tradeoffs, and supporting context that can be checked against your simulator before anything reaches the plant.
What you can optimize
Start with the business objective, then let the model search within real process constraints. Profitability, throughput, emissions, water use, and controllability can all be part of the same optimization problem.
Uphold safety
Every recommendation is constrained by process limits and safety parameters, so optimization stays grounded in the way your plant actually operates.
Improve profitability
We optimize against your economics, not generic objectives, helping teams improve throughput, yield, and operating margin in the same workflow.
Environmental impact
Energy use, emissions, water, and waste can all be included in the objective function so improvements support both performance and sustainability goals.
Flexible objectives
GHG reduction, controllability, spec quality, or another site-specific target. You define the objective and the model follows your priorities.
Workforce efficiency
Optimization runs become repeatable and easier to share, giving engineers more time for high-value decisions instead of manual scenario hunting.
Why teams trust us
Chemical engineering, applied AI, and industry advising in one team, so recommendations are technically grounded and practical to implement.
Henry Diehl
Co-Founder & CEO
Michael Perry
Co-Founder & CTO
Robert Wygle, PE
Advisor
Micah Green, PhD
Advisor
Jack Rodden, MBA
Advisor
We combine process intuition with modern machine learning so optimization is easier to trust, explain, and maintain after the engagement ends.
Talk through a first project
In a 30-minute intro, we can review your process, constraints, and economics to see whether AI optimization is a fit and what a practical first project could look like.
- Improve yield, margin, or throughput on an existing unit
- Reduce energy, emissions, water use, or waste within constraints
- Build a digital twin your team can keep using after the project
Bring a specific unit, plant objective, or data problem. We will help you assess likely value, rough scope, and the right first step.
