At Flux, we’re building an AI-native hardware design tool that gets better the more you use it. This month, we’re rolling out major upgrades to Copilot’s reasoning, transparency, and layout performance—plus some crucial fixes and a big leap in modular design reliability.
Let’s get into it.
🧠 Copilot just got a lot smarter
A few weeks ago, we launched Copilot Knowledge—a way for you to teach Copilot how you work. As you approve suggestions, they become part of your personal or project memory: vendor preferences, naming conventions, design rules, review checklists, and more.
The result? Faster decisions, fewer mistakes, and better suggestions without repeating yourself.
Now we’re introducing the next layer: System Knowledge. This is a shared, curated knowledge base built by our team of senior engineers—including folks from NASA and other top hardware teams. They’re constantly encoding real-world insights, best practices, and edge cases that help Copilot make smarter choices across the board.
By combining your personal knowledge with system-wide expertise, Copilot becomes more responsive, more accurate, and more relevant to how modern hardware actually gets built. It now blends tribal wisdom with engineering logic to solve problems that used to require manual intervention—or years of experience.
Want to shape what Copilot knows? Hit us up on Canny with ideas—naming conventions, routing strategies, design rules. The more we share, the better it gets for everyone.
📦 Copilot now shows its work
No more black box. Copilot now gives real-time feedback as it processes part selections and schematic requests.
You’ll see a live execution trace—how it searches, filters, ranks, and substitutes. It’s like watching a checklist unfold in real time. Whether you’re validating a part or building a schematic from scratch, Copilot shows exactly what it’s doing, and why.
⚡ Speed and precision upgrades
Alongside transparency, we’ve made core performance improvements to Copilot’s part-selection engine:
- 2–3× faster library searches
- More accurate MPN targeting
- More consistent behavior when adding parts in bulk
These changes reduce wait time, improve reliability, and keep your momentum flowing during schematic design.
🔝 Auto-Layout improvements
Thanks to your feedback, AI Auto-Layout just got a serious upgrade:
- Supports your full via config, including microvias
- Avoids awkward trace angles
- Smarter pad exit direction logic
- Better handling for castellated edge pins
These changes make Auto-Layout results cleaner, more production-ready, and easier to tweak.
🔜 Coming soon to Auto-Layout
We’re not done yet. Here’s what’s coming next:
- Differential pair routing—even when pins aren’t adjacent
- Smarter net spacing when layout allows
- Routing prioritization based on net type (power, clock, etc.)
🧱 Module reliability fixes
We resolved a core issue where polygons broke inside modules—restoring full support for airwire and copper DRC checks in modular designs. Now your modules behave predictably from design to fab.
Try it out
All of this is live now—log in to Flux and give the latest updates a spin. As always, keep the feedback coming. We’re building Flux Copilot with you.
More in AI & Prompt Engineering for Hardware
Let’s start building.
Let’s build your idea together. All you need to do is describe what you want to make, or what problem you want to solve. Flux will work with you to make it a reality.