Design Circuits with Natural Language: Copilot Upgrade

Ever since we launched Copilot, our goal has been to create a truly collaborative design partner—one that goes beyond offering advice to actually executing on your behalf. Today, we’re thrilled to unveil the next major leap in that journey!

Copilot is now powered by more advanced reasoning models and has the full context of your project—datasheets and your custom design rules—enabling far faster, and more accurate recommendations than any standalone AI chatbot could offer. Best of all, you remain in control, free to accept or refine any of its actions as you progress.

This isn’t just an upgrade—it’s a whole new way to design hardware with AI. Let’s dive in!

Get Started Easier & Generate a BOM

Starting a new hardware project can feel overwhelming. Even if you know exactly what you want to build, transforming your idea into a functional schematic requires choosing components, verifying specs, and ensuring each piece fits together seamlessly.

The completely overhauled Copilot takes the guesswork out of this process by engaging you in a focused conversation about your project. Instead of juggling datasheets and sourcing websites, you simply outline your goals—like “I want to build a battery-powered sensor module with an ESP32”—and Copilot follows up with questions to refine your requirements as a seasoned hardware engineer would.

Copilot: "Sounds great! Which sensors would you like to use—temperature, humidity, or something else?"

Once your specs are clear, it compiles a Bill of Materials (BoM) in seconds, giving you a curated list of compatible, optimized parts. You can review and adjust any element of the BoM before moving forward, confident that you’re building on a rock-solid foundation—without wasting hours on manual research.

Add Components Directly to the Schematic

After finalizing your BoM, Copilot can take the next step by placing each component into your schematic—whether it’s a microcontroller, sensor, or power module. It searches the library and drops parts onto your canvas in a way that avoids overlap, so you’re not starting from scratch.

Decoupling Capacitors in Seconds

With your key components in place, the next step is fine-tuning the design—often starting with decoupling capacitors. Traditionally, you’d consult datasheets for recommended values and place each capacitor manually. Copilot streamlines all of that into a single prompt:

You: "List all the decoupling capacitors that are needed for this design."

Copilot: "Below is a recommended list of decoupling capacitors for your sensor module design. Would you like me to add them to your schematic?"

Copilot identifies which parts need decoupling, calculates the optimal values, and places the capacitors next to their relevant pins. This alone can save you hours of research and tedious work, letting you maintain focus on the bigger picture.

Automatically wire components

Once your parts are positioned, the next challenge is ensuring everything is correctly wired—pin assignments, best practices, net labeling, and so forth. With Copilot, wiring becomes as straightforward as telling it which components to connect:

You: "Can you connect @U1 to @U3?"

Normally, this would mean double-checking pinouts, referencing datasheets, and verifying each signal. Copilot handles those checks behind the scenes, making the required connections in seconds. You can still review and adjust as needed, but the bulk of the manual labor is eliminated.

Smart Component Replacement

Even if your schematic is fully functional, there’s often room for optimization—maybe you need a sensor with higher resolution or a regulator that handles more current. Instead of searching for alternatives on your own, let Copilot do the legwork:

You: "Find a better alternative for U1 in the @library that has a higher resolution."

Copilot scans for a suitable match, checks it against your existing circuit constraints, and offers a drop-in replacement. Once confirmed, it updates your schematic accordingly—no library lookups needed.

More Ways to Guide Copilot

While Copilot automatically picks the best tools and models for most tasks, sometimes you need a little extra control. Whether you’re pulling equations from a datasheet, sourcing components from the Flux library, or performing quick calculations, prepending an "@" symbol tells Copilot exactly which tool or model to use—so you can fine-tune your workflow.

Tool-Specific Commands

• @file ties Copilot directly to any files attached to your project—like datasheets or design briefs—ensuring it references the most relevant information.
• @library gives Copilot a direct line to the entire Flux component library, making it easy to find, compare, and insert compatible parts.
• @calculator lets Copilot perform real-time math using datasheet equations, so you can size components accurately without manual computations.

Model Selection Made Simple

You can also choose from multiple AI models, each suited to a specific task:

• Advanced Reasoning – More capable but slightly slower, ideal for deeper technical analysis.
• General – A balanced option, both accurate and responsive for everyday tasks.
• Speedy – Faster responses for quick checks, though less detailed.

This flexibility ensures you’re always getting the right balance of speed and detail.

✨ Pro Tips for Best Results

• ‍Break Tasks Into Steps. Copilot works best when given one task at a time. Instead of asking it to build an entire schematic, begin by telling Copilot what you want to build, then give it specific prompts like “@copilot, suggest a temperature sensor suitable for outdoor environments.” After selecting a component, proceed with, “@copilot, add the recommended temperature sensor to the schematic.” Finally, instruct, “@copilot, connect the sensor's output to the analog input of @U1.”​‍
• Provide Context. The more details you add to your project—like goals, requirements, attached datasheets, or components on the canvas—the better Copilot performs. These details act as Copilot’s “long-term memory” across conversations.‍
• Use Specific Tools. Utilize tags like “@library” to reference parts libraries or “@file” to attach datasheets. Specifying design parameters further refines Copilot's recommendations.​‍
• Watch for Action Buttons. When Copilot identifies executable tasks, an action button will appear in the chat interface. Clicking this allows Copilot to perform the task directly on your design. If the button doesn't appear, consider rephrasing your request or breaking it down into smaller steps.

AI as Your Design Partner

This release marks a pivotal step toward the future of AI-assisted engineering. Flux Copilot isn’t just here to answer questions—it’s evolving into a genuine design partner, one that helps you move faster, reduces friction, and keeps you focused on real engineering challenges. But we’re only scratching the surface of what’s possible.

Our larger vision is to automate even more of the tedious steps in hardware design—like routing and board layout. If you haven’t already, check out our AI Auto-Layout feature, which aims to take care of basic board routing so you can iterate faster and get to market sooner.

Join the Beta & Share Your Feedback

Right now, Copilot is in a community beta. While it’s more capable than ever, it’s not perfect—and that’s where you come in. Your feedback will shape its evolution. Tell us what works, what doesn’t, and what actions you’d like Copilot to tackle next.

We’re rolling out new features in the coming weeks. If you don’t see them yet, keep an eye on Flux—or join our Slack and message Nico to get in early.

Ready to experience AI that truly takes action?

{{open-flux-try-copilot}}