CircuitLab

CircuitLab AI helps electronics students, Arduino/ESP32 makers, and laboratory users investigate low-voltage circuits through a structured, evidence-driven workflow. Circuit photos, serial logs, and measurements often live in separate places, making it difficult to decide what to check next or determine whether the expected behavior has been observed. Users describe the circuit, confirm its connections, define measurable criteria, and attach evidence. Muse Spark analyzes the available context, compares possible explanations, and proposes a targeted next check. As the operator records observations, CircuitLab updates the investigation and preserves its history. During the hackathon, we built the bilingual investigation workflow, Meta Spark integration with a session-scoped observation tool, browser-based USB monitoring, and backend validation and reporting. The application supports web and Android, with compatible USB connections handled through a supported desktop browser. The implementation separates model proposals from backend validation. Typed observations preserve their source, PostgreSQL stores session history, and a durable worker processes analysis jobs. Completion checks require fresh observations against the declared criteria and a separately recorded functional test. Manual entries, board-reported telemetry, and instrument readings retain distinct provenance. Our hardware demo follows an Arduino Uno project with RFID, temperature, humidity, and GPS modules through test preparation, serial evidence capture, investigation of a missing GPS fix, and session reporting. The operator records observations and the functional-test outcome alongside the earlier messages. CircuitLab applies Meta’s models to understanding physical electronic systems. Its potential value is reducing trial-and-error troubleshooting and making investigations easier to review, reproduce, and continue. Project website in English: https://circuitlab.doc-ia.cloud/en/ Web application: https://circuitlab.doc-ia.cloud/app/