KiCon Europe 2025

Large Language Models (LLMs) and Vision Language Models (VLMs) are transforming software development, but their application in hardware design is still emerging. To be genuinely useful in electronics design, it requires a structured understanding of the project. This talk presents AmpereBrain, a proof-of-concept that gives a local AI the tools to read and interpret KiCad projects in a structured way. We demonstrate how providing the AI with access to project specifications, schematic diagrams, and component-level details enables it to become a practical assistant. This allows for reliable, AI-powered help with tasks like documentation, component queries, and basic schematic reviews, laying the groundwork for more advanced, structured design modifications.

This talk introduces a new, "dead-simple" paradigm for KiCad libraries, powered by AmpereBrain. We will demonstrate a workflow where managing complex, version-controlled libraries becomes as easy as dragging and dropping a file. AmpereBrain automates the entire lifecycle: it ingests component files, organizes them into a Git-backed repository, and automatically searches for, downloads, and processes datasheets. We will showcase how this system creates an intelligent, queryable component database by loading datasheet information into a Large Language Model (LLM), allowing you to ask complex questions in natural language.

With ngspice integration in KiCad, simulations have never been more accessible in KiCad. But when the design gets complex, managing the simulations is complex. This talk presents a new methodology, enabled by AmpereBrain, that establishes a single, unified schematic as the source of truth for both design and multiple simulation scenarios. We will demonstrate a workflow centered on logically grouping components within your schematic. By programmatically activating these groups, you can define precise simulation boundaries on the fly, while the tool automatically filters out components irrelevant to the simulation. This structured approach is the key to enabling robust, repeatable simulation and paving the way for a true CI/CD workflow for hardware development.