IPC International Community magazine an association member publication
Issue link: https://iconnect007.uberflip.com/i/1543307
120 I-CONNECT007 MAGAZINE I FEBRUARY 2026 problem solving. In an ideal workflow, they spend their time on the decisions that shape the product, while technology handles the rest. AI systems have tremendous value because, rather than replacing engineering judgment, they handle first-pass review work, surfacing potential risks early and directing attention to what matters most. As one example of this approach, we've built DRCY, an AI-powered agent designed specifically for hardware design. The Power of Native Data Access A hardware-specific AI differs fundamentally from a general-purpose AI solution for design analysis. When you export a schematic to PDF or take a screenshot for an AI chatbot, you throw away most of the information that matters. An image shows shapes and text, but it does not necessarily under- stand that a specific line is net VCC_3V3, that it connects to pin 7 of U4, or that U4 is a voltage regulator with specific input/output requirements. Compound that with more complex ECAD data objects—busses, multi-channel sheet hierarchy, and design rules— and you end up with an ECAD- specific logic to reason out before an agent can start analyzing design intent. That adaptor layer is where AllSpice specializes. We built native design data access into DRCY. It sees nets, components, properties, hierarchy, and connectivity, with a schema specifically designed for AI reasoning. It works natively with all ECAD formats. This native access allows it to catch subtle issues that might otherwise be missed until later stages. How AI Reasoning Changes Design Review Agentic AI is a term that is often used, but what does it mean in practice? For DRCY, it means that rather than running a fixed analysis pipeline, it decides what to investigate based on what it finds in your design. For example, when DRCY encounters a volt- age regulator, it doesn't just check that the part exists. It will pull the datasheet to ver- ify input/output voltage specs, trace the input net to confirm the source voltage is within range, check that output capaci- tors meet the recommended values, ver- ify the enable pin is properly controlled, and more. The path it takes depends on your specific design, not just a hardcoded checklist. We call these "skills": specialized tools a hardware-specific AI reviewer can invoke when needed. Think of it like hav- ing access to a well-organized reference library. DRCY can pull the right datasheet, cross-reference component specifications, trace signal connectivity, or apply domain- specific analysis heuristics: whatever the situation requires. Figure 1: A sample comment that DRCY posts in the AllSpice platform. " AllSpice.io set out to automate the repeatable parts of hardware design: the well-defined, rules- based tasks that make sure nothing gets missed and everything lines up."