CELUS Introduces AI Agent Workflow for Electronics Design Architecture

CELUS has announced the launch of an agentic workflow for its Design Assistant, a tool designed to help hardware engineers plan architecture and component selection at the earliest stages of electronics design. According to the company, the new workflow aims to reduce manual data sheet searches and lower revision rounds by structuring early planning before any schematic capture takes place.

Key Features of the Agentic Workflow

• Intent Capture: Engineers describe the desired product, while the assistant prompts for specifications like interfaces, power budgets, costs, and compliance, compiling these into a step-by-step plan.
• Human-in-the-Loop Approval: The assistant does not execute tasks automatically; engineers must review and approve the plan before execution and retain the ability to stop or edit the process.
• Architecture Generation: The system produces a block diagram and researches suitable components, providing direct citations to underlying datasheets.
• Self-Review and Alternatives: The tool checks designs against initial requirements, flags discrepancies, and ranks substitute parts based on cost, lead time, or preferred suppliers.
• Tool Integration: Outputs can be exported to standard industry tools such as Altium Designer, KiCad, Siemens EDA, and Autodesk Fusion.

Practical Implications and Verification

The reported workflow addresses common challenges in electronics engineering, such as managing compressed schedules and component shortages early in a project cycle. However, users should note that multi-step agentic requests take longer to process than traditional single-prompt queries as the system plans and verifies its work.

Engineering teams interested in evaluating the beta release can access it directly through the CELUS platform at app.celus.io. Independent verification of platform capabilities, compatibility, and component library accuracy can be conducted by testing sample projects against established EDA toolchains and comparing AI-suggested part availability directly with manufacturer datasheets.