The Flux Method

1. Start With the Outcome, Not the Tools

Describe what you want to build, not how to draw it.

Traditional CAD workflows begin with a blank canvas. You select components, draw schematics, model geometry, route traces, and configure countless settings before you have something that resembles your original idea.

But your goal was never to draw a schematic or model an enclosure.

Your goal was to build something that works.

With the Flux Method, you start by describing the product you want to create.

For example:

Design a battery-powered air quality monitor with a CO₂ sensor, a small display, Wi-Fi connectivity, and a 3D-printable enclosure. It should run for at least 24 hours on a charge and cost less than $50 to manufacture.

Flux can help translate that goal into an engineering design, identify requirements and constraints, evaluate tradeoffs, and begin implementation.

You don’t need to specify every component, connection, or dimension up front.

Start with your intent. Refine the details as you go.

Best practices:

• Describe the product’s purpose, functionality, and constraints.
• Include requirements like size, cost, battery life, and manufacturing method.
• Provide reference designs, datasheets, sketches, or existing CAD files when available.
• Let Flux propose an approach before diving into implementation.
• Focus on what the product needs to accomplish rather than how to operate the design tools.

2. Delegate the Work, Not Just the Questions

Treat AI as an engineer, not a search engine.

The first generation of AI design assistants answered questions and offered suggestions.

That’s useful, but it still left humans responsible for almost all the execution.

An AI-first workflow is different.

Instead of asking how to accomplish a task, ask Flux to accomplish it.

Rather than:

How do I connect this sensor to my microcontroller?

Try:

Add this sensor to my design, connect it to the microcontroller over I²C, and make sure the power supply can support it.

Rather than:

How should I design an enclosure for this PCB?

Try:

Design an enclosure around this PCB with mounting bosses, USB-C access, ventilation, and a snap-fit lid.

Flux can work directly on engineering artifacts, rather than merely describing what you should do.

Your role shifts from performing every operation to directing, reviewing, and refining the work.

Best practices:

• Delegate complete engineering tasks rather than individual CAD operations.
• Provide clear goals and constraints instead of lengthy step-by-step instructions.
• Let Flux choose an implementation approach when the details aren’t critical.
• Review important design decisions and ask for alternatives when needed.
• Use direct editing when it is faster or when precise manual control is necessary.

The goal isn’t to eliminate human expertise. It’s to multiply what one person or team can accomplish.

Example project: Pico Macro Keyboard

3. Think in Products, Not Disciplines

Build the whole product, not just the PCB.

Hardware doesn’t exist in isolated engineering disciplines.

An enclosure affects the PCB layout. A component choice affects the firmware. Thermal requirements affect mechanical design. Manufacturing constraints influence almost every engineering decision.

Yet traditional engineering tools separate these interconnected problems into different applications and workflows.

The Flux Method takes a product-first approach.

Instead of treating electrical design, mechanical design, firmware, and manufacturing as separate projects, treat them as different aspects of the same product.

One AI hardware engineer can work across these domains, keeping the full product context in view.

For example:

Replace the display with a larger model, update the PCB layout and enclosure to fit it, and adjust the firmware accordingly.

What traditionally required several engineers working across multiple tools can increasingly be coordinated through one conversation.

Best practices:

• Define requirements at the product level rather than only at the component level.
• Consider electronics, mechanics, firmware, and manufacturing together.
• Ask Flux to identify dependencies when making design changes.
• Evaluate how changes in one discipline affect the others.
• Keep the complete product, not just individual design files, as the source of context.

The unit of work is the product, not the CAD file.

See it in Flux: Flux for Teams

4. Iterate at the Speed of Ideas

Explore alternatives before committing to a design.

Traditional hardware development makes iteration expensive.

Changing a component might require redesigning part of a schematic, rerouting a PCB, modifying an enclosure, and updating documentation.

The effort involved encourages engineers to commit to decisions early, even when uncertainty remains.

AI makes it practical to explore more possibilities.

Instead of spending hours manually implementing an alternative, ask Flux to investigate it.

For example:

Compare three approaches to powering this device. Optimize for battery life, cost, and component availability. Recommend the best option and explain the tradeoffs.

Or:

Explore how much smaller we could make this product if we switched to a four-layer PCB and a different battery.

The goal is to make engineering exploration cheap enough that you can evaluate ideas before committing significant time and resources.

Best practices:

• Explore multiple approaches before settling on an architecture.
• Ask Flux to explain the tradeoffs between alternatives.
• Use simulations and engineering checks to evaluate designs.
• Challenge assumptions when requirements change.
• Iterate until the design meets your goals, not merely until it looks complete.

The faster you can explore and evaluate alternatives, the better your final product can become.

5. Verify, Don’t Just Trust

Use AI to do the work, but demand evidence that it works.

A plausible-looking design isn’t necessarily a working design.

Hardware must obey the laws of physics. Components have electrical limits. Mechanical parts must fit together. Manufacturing processes have tolerances. Firmware must interact with real devices.

That’s why verification is fundamental to the Flux Method.

Don’t just ask Flux to create a design. Ask it to evaluate that design against your requirements.

For example:

Check the power architecture, simulate the critical circuits, verify mechanical clearances, and identify potential manufacturing issues.

Or:

Review this design as if you were preparing it for production. What could go wrong?

AI can help perform engineering analysis, identify risks, and propose corrections, but the level of confidence must match the available evidence.

A successful simulation does not guarantee that a physical product will work.

Best practices:

• Define measurable acceptance criteria.
• Ask Flux to verify its work rather than simply report completion.
• Use simulations, electrical checks, and mechanical fit tests where appropriate.
• Examine assumptions and unresolved risks.
• Build and test physical prototypes to validate critical functionality.
• Apply additional independent review and formal verification where safety, reliability, or regulatory requirements demand it.

The goal isn’t just to generate designs faster.

It’s to build working hardware faster.

See it in Flux: Verification on Flux for Teams

6. Build Together, With AI

Make AI part of the engineering team.

Hardware development has always been collaborative.

But traditional CAD workflows often make collaboration difficult. Designs are scattered across files, applications, and computers. Important decisions get lost in messages and meetings.

With Flux, humans and AI can collaborate in a shared engineering environment.

Team members can review designs, contribute expertise, and direct work without each person needing to manually execute every task.

A mechanical engineer can help refine electronic packaging. An electrical engineer can evaluate mechanical constraints. A product designer can participate in decisions that previously required specialized CAD knowledge.

AI helps bridge the gaps between disciplines.

Best practices:

• Share projects early rather than waiting for formal design reviews.
• Keep design discussions connected to the actual engineering work.
• Bring stakeholders into the process before important decisions become expensive to change.
• Use AI to help explain design decisions across disciplines.
• Preserve human review and accountability for important engineering decisions.

AI doesn’t just make individual engineers more productive.

It enables smaller teams to accomplish work that previously required much larger, more specialized organizations.

7. Keep Moving From Idea to Manufacturing

The design isn’t finished until you can build it.

The purpose of hardware engineering isn’t to create CAD files.

It’s to create physical products.

Manufacturing should be considered throughout the design process, not as a final handoff.

Component availability, assembly methods, mechanical tolerances, testability, and production costs all influence whether a design can become a successful product.

The Flux Method encourages designing for real-world constraints from the beginning.

For example:

Prepare this product for manufacturing. Review component availability, identify assembly risks, and generate the documentation needed for fabrication and assembly.

Use AI to help connect design decisions with manufacturing requirements rather than treating manufacturing as somebody else’s problem.

Best practices:

• Consider manufacturability from the earliest design stages.
• Include target production volumes, cost constraints, and preferred manufacturing processes.
• Evaluate component availability and potential substitutions.
• Create fabrication, assembly, testing, and bring-up documentation.
• Validate the design with prototypes and incorporate what you learn into subsequent revisions.

The goal is to go from an idea to a real, working product with as little unnecessary friction as possible.

Example projects: Featured projects, Powerbank board