Reality → understanding → action
Fuse live operational data with first-principles models to estimate physical state, explain deviation and recommend safe action.
Observe the existing process
Infer physical state
Optimise operation
Generative industrial design · Powered by Veldar physics
Design from the
target state.
Alchemy turns engineering intent into physics-grounded process concepts. Brief what a system must achieve, set the constraints that matter, and explore designs that can be simulated, compared and refined before capital is committed.
Product vision · In development · Human engineering approval remains essential
Alchemy design workspace · study ALC-0241
01 · Refineing intent
Design a duty/standby pump system delivering 450 m³/h of process water across 42 m static head. Minimise lifecycle energy use, tolerate ±18% demand variation, and preserve operation during one-pump maintenance.
Confirm fluid temperature and density range
What is the allowable NPSH margin?
Prioritise lowest CapEx or lowest 15-year total cost?
Brief redundancy and site footprint constraints
02 · Developd system candidate
Feed
Tank
Tank
Duty
Pump
Pump
Process
Header
Header
Standby
Pump
Pump
Control
Valve
Valve
12.8%Energy reduction
99.4%Availability
3Viable concepts
Design study · Pump network · 24 feasible architectures generatedConstraint check · NPSH margin maintained across demand envelopeScenario · Fouled exchanger case simulated at year fiveOptimisation · Lifecycle energy reduced against reference designRefineing review · Two unresolved assumptions require confirmationDesign study · Pump network · 24 feasible architectures generatedConstraint check · NPSH margin maintained across demand envelopeScenario · Fouled exchanger case simulated at year fiveOptimisation · Lifecycle energy reduced against reference designRefineing review · Two unresolved assumptions require confirmation
One physics engine · Two directions
Sentinel understands what exists.
Alchemy designs what should exist.
Both products use the same physics knowledge, process context and engineering constraints. Sentinel begins with a live plant and works towards better decisions. Alchemy begins with a desired outcome and works backwards towards a viable process design.
⇄
Intent → physics → design
Translate objectives and constraints into candidate process architectures, evaluate them under realistic conditions and refine the strongest concepts.
Brief target state
Develop feasible designs
Test and compare
Guided generative engineering
From a plain-language brief
to an engineering-ready concept.
Generative design works best when goals, constraints and evaluation criteria are explicit. Alchemy structures that process, asks for missing information, explores the design space and keeps engineers in control of every consequential choice.
Brief
Describe the process, target output, operating envelope and business objective in familiar engineering language.
Question
Alchemy identifies missing assumptions, conflicts and unknowns, then requests the information required to proceed.
Develop
Create multiple system architectures, equipment arrangements and operating strategies that satisfy hard constraints.
Test
Run physics-based models across normal, transient, degraded and failure conditions—not just the nominal design point.
Compare
Rank candidates against energy, throughput, safety, resilience, emissions, footprint, CapEx and lifecycle cost.
Refine
Export assumptions, calculations, equipment requirements and an auditable basis of design for expert review.
Shared Veldar physics engine
Not a text generator.
A constrained engineering system.
Alchemy uses language to capture intent, but design generation is governed by first-principles models, equipment behaviour, operating envelopes and explicit constraints. The objective is not to produce persuasive answers—it is to produce traceable design candidates that survive engineering scrutiny.
Refineing intent
Targets, process description, feed conditions, required outputs, site constraints and preferred trade-offs.
System context
Assets, streams, unit operations, materials, interfaces and allowable relationships represented in a common model.
Physics primitives
Mass and energy balances, thermodynamics, fluid mechanics, heat transfer, reaction and equipment models.
Candidate development
Constraint-aware generation, optimisation and trade-space exploration across candidate architectures.
Engineering evidence package
Assumptions, equations, sensitivity, uncertainty, scenarios and design rationale prepared for human review.
Explore the trade space
Compare designs before steel is cut.
Traditional design often converges early on a familiar configuration. Alchemy keeps alternatives visible and evaluates how each behaves across the full operating envelope, including uncertainty, degradation and future demand.
CandidateEnergyCapExResilienceFootprint
A · Parallel VSD pumpsBestMediumHighMedium
B · Single duty + bypassMediumLowLowLow
C · Three-pump stagedHighHighBestHigh
D · Elevated storage bufferMediumHighHighHigh
Core capabilities
Design faster without making engineering invisible.
01
Conversational requirements capture
Turn an initial brief into structured requirements while exposing ambiguity before it becomes expensive rework.
02
Constraint-aware generation
Develop only within defined physical, regulatory, safety, spatial and commercial boundaries.
03
Multi-physics evaluation
Evaluate interactions between flow, pressure, heat, energy, materials and equipment rather than optimising components in isolation.
04
Scenario and sensitivity testing
Test feed variability, demand growth, degraded equipment, control failure and future operating cases before final selection.
05
Lifecycle optimisation
Balance capital cost against energy, maintenance, resilience, emissions and expected operating life.
06
Traceable engineering evidence
Keep assumptions, calculations, sources, model versions and decision rationale visible for review and governance.
Where Alchemy could begin
From a single system upgrade
to a new process concept.
The practical entry point is narrow: one design decision with clear requirements and measurable trade-offs. The same architecture can expand as the physics engine and validated model library grow.
Application 01
Pump and fluid systems
Size and configure pumps, pipework, valves, storage and controls against variable demand, energy and reliability objectives.
Design questionBest system—not just best pump
Application 02
Brownfield upgrades
Evaluate retrofit options against existing equipment, space, utilities and production constraints before shutdown scope is fixed.
Design questionWhat change creates most value?
Application 03
Process-unit concepts
Explore flowsheets, equipment trains and operating strategies for separation, heat transfer, treatment and materials processing.
Design questionWhich architecture meets the target?
Application 04
Capacity and debottlenecking
Develop alternatives for additional throughput while respecting upstream, downstream, utility and equipment constraints.
Design questionWhere should capital be placed?
Application 05
Greenfield process design
Move from a target product and operating envelope towards candidate process architectures and a defensible concept-design basis.
Long-term visionFrom intent to process
Refineing governance
Generative does not mean ungoverned.
Alchemy is conceived as an engineering co-design environment, not an autonomous sign-off authority. Every design should make its assumptions, limitations and unresolved questions clearer—not hide them behind an answer.
Engineer approval
Qualified engineers review requirements, model selection, constraints and final design decisions.
Explicit uncertainty
Unknown inputs and model confidence remain visible, with sensitivity shown where uncertainty affects selection.
Code and standard checks
Applicable standards and organisational rules become constraints, with exceptions surfaced for review.
Auditable rationale
Every recommendation links back to requirements, assumptions, calculations and evaluated alternatives.
Shape the product vision
Bring us a design problem worth rethinking.
Alchemy is an emerging product concept built on the same physics engine as Sentinel OS. We are looking for engineering and operations teams willing to test where generative, physics-grounded design could reduce iteration time, expand the option space or improve lifecycle outcomes.
- Start with one bounded system
- Use real engineering constraints
- Compare against the current workflow
- Validate with qualified engineers
- Turn learning into reusable models
Questions
Alchemy FAQ
No. Alchemy is intended to accelerate requirements capture, option generation, simulation and trade-space evaluation. Refineing judgement, discipline expertise, code compliance and formal design approval remain essential.
Conventional tools are powerful once an engineer has already defined the configuration to model. Alchemy aims to begin earlier: structuring intent and constraints, generating candidate process architectures, orchestrating the appropriate physics models and comparing alternatives. It should complement established simulation, CAD and engineering tools rather than replace them.
That is a long-term vision, not a credible starting claim. Early use cases should be bounded systems or unit operations with well-defined requirements, validated models and available engineering data. Capability can expand progressively as the model library and evidence base mature.
Sentinel uses live data and physics to understand and optimise operating systems. Alchemy uses the same physics foundation in reverse: beginning with a target state and constraints, then generating and evaluating designs capable of achieving it. Operational evidence from Sentinel can ultimately improve future Alchemy designs.
A focused study around one system—such as a pump network, heat-transfer loop or brownfield capacity upgrade. Together we would define requirements, capture constraints, reproduce the current engineering baseline, generate alternatives and compare the quality, speed and transparency of the resulting workflow.