An AI that thinks like a PCB engineer, placing components, spotting errors, and ensuring precision, while keeping full human control at the centre. How is that possible?
An Indian deep tech startup is attempting to bring context-aware intelligence into one of hardware engineering’s most manual workflows. AutoCuro, founded by Manav Marwah, has developed an AI-assisted printed-circuit (PCB) design automation engine that operates within established EDA environments, focusing on schematic interpretation, component placement, and baseline routing.
The system integrates with platforms such as KiCad, Cadence Design Systems, and Altium Designer, allowing engineers to continue using familiar design tools. Instead of treating every connection as a generic net, the engine is designed to interpret the functional role and electrical context of components before generating placement suggestions. “The goal is to replicate human-level reasoning. While it may not yet reach a human engineer’s peak optimisation, it significantly reduces design time while maintaining quality,” added Manav.
For instance, if a schematic includes a crystal oscillator and a microcontroller, the AI evaluates the proximity requirements for timing accuracy. If it encounters decoupling capacitors, it considers their relationship to power pins and supply lines. These are decisions traditionally guided by engineering judgment. The software parses schematic folders, footprints, and board constraints, then proposes placement within defined outlines and mechanical limits. Routing is generated with awareness of layer stack configuration, power distribution, and controlled signal paths.
Unlike static auto placement tools, the system can respond to change. When a component is moved or a new part is introduced, the schematic context is reinterpreted, and placement relationships are recalculated before the layout is updated. Instead of freezing the design into a fixed arrangement, it continuously evaluates connectivity, hierarchy, and constraints as the board evolves. Engineers can also review each revision, but the underlying structure adjusts with the design rather than forcing teams to restart placement decisions from scratch.
“Our focus is on intelligent placement and efficiency, not necessarily minimising board size. It ensures components are placed optimally according to electrical context, which is critical for performance and signal integrity,” added Manav.
Once placement and routing are complete, engineers can proceed with standard verification flows. The engine also generates a detailed diagnostic report flagging routing conflicts, design rule check (DRC) violations, and potential design for manufacturing (DFM) and design for assembly (DFA) concerns, with a portion of these addressed before the design reaches human review. More complex routing paths and edge-case scenarios continue to require supervision. Final validation and sign-off remain firmly with the engineer, reinforcing the assistant‑led model rather than fully autonomous design. It is clear that the AI assistant is here to help engineers, not replace them.
