What if you could sketch a circuit idea on a napkin or describe it in plain language, and an AI instantly transformed it into a working schematic? This once-futuristic notion is becoming reality as artificial intelligence finds its way into electronics design workflows. In an industry where time-to-market and design complexity are growing concerns, AI-driven tools promise to automate tedious tasks and amplify engineers’ capabilities. But how close are we to that vision, and what does it mean for professionals working on mass-produced electronics?
AI + PCB Design in the Wild
Electronics engineers in the open-source community have been busy exploring AI’s potential in PCB design. Take the example of a few KiCad MCP server projects, which are specialized servers that provide context (MCP stands for Model Context Protocol) and allow large language models (LLMs) drive the KiCad EDA software through natural language commands. In practice, this means an AI agent (like ChatGPT or Claude) could “manage KiCad projects, design PCBs, place components, route traces, and generate outputs” just by interpreting a user’s requests. Other ambitious projects go a step further: using multiple AI agents to plan a circuit, pick parts, write code (using for example SKiDL, a Python PCB design language), and even debug errors in a reasoning loop. The fact that such tools exist and people are investing their times on such projects already means a lot.
Despite the enthusiasm, these efforts face notable limitations. The AI often needs heavy oversight, and results can be hit-or-miss. A major reason is scale: compared to the billions poured into AI coding assistants, relatively little investment has gone into PCB design automation, and this is, among other factors, because the software industry has a larger market value for now. In short, the potential is real: natural language schematics, PCB routing, but the polish and reliability are not quite there, and there is a lot to be done.
Inspiration from Software Development: AI Assistants that get you
While electronics AI is just emerging, software development workflows have been revolutionized by AI over the past two years, setting a benchmark for usability. Developers now have AI copilots like Cursor, Lovable, and V0 that drastically streamline coding and app design, and these successes are starting to shape expectations for electronics design.
Cursor, for instance, is an AI-infused code editor that makes interacting with an LLM feel like a natural part of writing software. Instead of switching to a separate chat window, Cursor brings the AI right into your IDE. You can highlight a block of code and ask, “Can you optimize this?”, and the AI suggests edits in place. There’s no clunky copy-paste dance. The model already has full context of your codebase and responds within your workflow. The real magic of Cursor isn’t a more powerful model, but a more seamless user experience: the AI feels like a partner embedded in your environment, ready to help without being explicitly invoked each time. This kind of fluid, context-aware interaction is a UX breakthrough, one that lets engineers focus on intent (“make this faster” or “explain this”) rather than micro-managing the AI with perfect prompts.
Meanwhile, AI-driven app builders like V0 and Lovable have shown how far “user intent recognition” can go in software design. Instead of manually coding a UI or setting up a database, you simply describe the application you want and let the AI do the heavy lifting. For example, “Dashboard with a sidebar and table of user data” is all you need to tell, and it will instantly generate a clean frontend with a styled layout matching that description. These tools infer what you are trying to achieve and produce high-quality results without making you struggle with low-level details. And you can always give feedback to the agent and iterate over the initial results to fine-tune your outputs. The UX is natural and goal-focused, and it raises the bar on what we expect from our design software.
It is not hard to imagine why hardware engineers are watching these developments closely. If AI can translate a sentence into a functional web app or optimize code with a quick comment, why shouldn’t our PCB tools let us describe a circuit and get a generated design? The success of AI in software is creating a vision for electronics design: more intent-driven interfaces, fewer manual tasks, and an experience that feels like discussing your ideas with an expert colleague, not fighting CAD software.
Towards Intent-Driven Electronics Design
Bringing that vision into the electronics world comes with unique challenges: physical laws and component details are less forgiving than a mistyped line of code, but the trajectory is clear. We are already seeing the first generation of AI-assisted electronics design platforms, and they are evolving fast. These platforms are taking inspiration from the software side (rich context, natural language UX) while also addressing the reliability and domain-specific accuracy that hardware demands.
One approach gaining traction is to mix human-guided structure with AI automation, or, in other words, let the engineer outline the “what” and have the AI figure out the “how.” This is where the concept of a high-level functional block diagram comes in. Rather than starting a design by drawing schematics and wiring up pins (the “maze” of details), you start by mapping out the functional blocks (“the map”) of your system: e.g. an MCU here, a power supply there, a sensor, a wireless module, and so on. That diagram helps capturing your design intent, what building blocks the design needs and how they connect, without yet specifying every resistor and capacitor.
AI-Powered Design Assistants: From Blocks to BOM
Enter CELUS Design Platform, our effort to realize exactly this idea. It takes a map-first approach: using AI to help you define the system you need to build, understanding your design intent, and then using this knowledge to recommend the best solutions for you. Here’s how it works in practice:
1. Describe your intent. You can drag-and-drop blocks like “Microcontroller” and “Battery Charger” onto a canvas, define their interfaces (UART, I²C, etc.), or even just type a quick description and let the AI generate the block diagram for you. In fact, our Design Assistant allows you to simply write what you want or upload a rough sketch, and it will populate the canvas with the appropriate functional blocks.
2. Component search and recommendations. Once the functional blocks are in place, the platform searches through a vast component knowledge base to find parts and circuits (CUBOs) that fit each block’s requirements. In other words, if your block diagram says you need a 3.3V regulator or an IMU sensor, the platform will shortlist specific CUBOs for those functions, considering your constraints. This isn’t a blind guesswork approach: we leverage a database of components and CUBOs to ensure the recommendations make sense.
3. Generation of schematics and BOM. Now comes the payoff: with one click, our platform transforms that abstract block diagram into a complete, wired schematic and Bill of Materials. The system pulls in the chosen components for each block, connects the nets according to the architecture, and produces a full electronic schematic ready for export to your favorite ECAD tool. In minutes, you get what could normally take weeks of part research and schematic drafting, a design that is logically consistent and backed by real component data.
The CELUS Design Assistant can even start from a simple hand-drawn sketch. In the example above, an engineer’s sketch diagram is fed into the platform. The AI interprets the rough blocks and automatically creates a tidy functional block diagram on the canvas.
This AI-assisted workflow doesn’t remove the engineer from the loop, it augments the engineer. You still verify the suggestions, make adjustments, and apply your domain knowledge to fine-tune the design. The difference is you spend your time on high-value decisions (choosing between topology A or B, trade-offs between component cost vs. performance) rather than other tasks like checking distributors’ websites or drawing lines repetitively.
What’s Next?
The advent of AI in electronics design is just beginning. The open-source community gives us a taste of what is possible, and the software development world has shown us what “good” looks like in terms of fluid, intent-driven workflows. Now platforms like CELUS are bridging those insights into our daily engineering tools, automating the tedious parts of schematic capture and component selection while letting us focus on the creative and critical thinking aspects of design. It is a bold new direction, and also one that invites collaboration between engineers and these AI tools in a whole new way.
So, what AI-powered features or user experiences would make your electronics design workflow smoother? We want to hear from the real experts: electronics engineers on the front lines. Would you like agent-driven adjustments on CUBOs? Auto-generated test plans? Real-time cost and supply chain optimization as you design? By voicing your needs and ideas, you can help shape the next generation of AI-driven design platforms.
