PCBRouteAgent: How Close Are LLM Agents to Becoming PCB Routing Engineers?
Abstract
Automating printed circuit board (PCB) routing requires coordinating board-level connectivity decisions with geometric execution under strict design rules. Although recent agents can interact with native routing tools, reliable routing on complex boards remains challenging. We present PCBRouteAgent, a multimodal PCB routing agent inspired by how human engineers coordinate global strategy with local execution. A Global Routing Planner combines board images and structured electrical and geometric information to specify connection topology, routing order, corridors, and layer transitions. A Detailed Routing Executor realizes these guides through KiCad's native routing engine and performs bounded local recovery. Unresolved failures are returned as structured feedback, enabling targeted guide revisions while preserving accepted routes. We evaluate PCBRouteAgent on 247 synthetic and real-world boards from four PCBWorld-Bench test splits. It achieves clean-pass rates of 100%, 94.9%, 100%, and 40% on D2, D3-A, D3-B, and D3-C, respectively. On the three real-world subsets, PCBRouteAgent outperforms the best evaluated baseline in clean-pass rate by 13.9, 36, and 10 percentage points, respectively.
Community
🚀 Can AI agents automate PCB routing?
PCB routing is a critical yet time-consuming step in electronic design, traditionally requiring substantial engineering effort and specialized expertise.
We introduce PCBRouteAgent, an AI agent framework for automated PCB routing.
Our key results:
- 🏆 Achieves up to 100% clean-pass rate on evaluated benchmark splits.
- 📈 Improves performance by up to 36 percentage points over the strongest evaluated baselines on real-world subsets.
- 🔧 Demonstrates the potential of AI agents to tackle complex, real-world engineering workflows.
We hope this work helps bridge the gap between general-purpose AI agents and practical electronic design automation (EDA).
🌐 Project: https://chip-ai-agent.github.io/research/pcbrouteagent/
💻 Code: https://github.com/chip-ai-agent/PCBRouteAgent
We'd love to hear your thoughts and feedback!
I love how honest this abstract is: splitting the job into a global planner and a detailed executor is exactly how I'd want a routing agent to think. Where does it still fail on complex boards, the planning or the execution?
Yes—global planning is still a bottleneck: early routing choices can block later connections, and replanning doesn’t always resolve the congestion. The executor handles local geometry well, but can’t fully compensate for a poor global plan.
Where's the model tho, sidebar says no models cite this paper and I looked the author profile, don't see there either.
PCBRouteAgent is a routing framework, not a newly trained model. It uses existing LLMs, so there are no model weights to download. Code: https://github.com/chip-ai-agent/PCBRouteAgent
PCBRouteAgent is a routing framework, not a newly trained model. It uses existing LLMs, so there are no model weights to download. Code: https://github.com/chip-ai-agent/PCBRouteAgent
This is an automated message from the Librarian Bot. I found the following papers similar to this paper.
The following papers were recommended by the Semantic Scholar API
- Validity-Preserving Hierarchical RL for Joint Routing and Switch Placement in EDA (2026)
- LAMP: LLM-Driven Analog Placement through Adaptive Tool Composition (2026)
- LATCH: LLM-Assisted Timing Closure Harness for Post-Route FPGA Optimisation (2026)
- EngiWorld: What Can Frontier Agents Deliver in Professional Engineering Environments? (2026)
- Whose turn is it? Context-guided model routing for stateful agents (2026)
- Demand-Driven Routing Space for Lattice-Surgery Compilation (2026)
- APPlace: Co-Optimizing Device Locations and Access Points for Routing-Aware Analog Placement (2026)
Please give a thumbs up to this comment if you found it helpful!
If you want recommendations for any Paper on Hugging Face checkout this Space
You can directly ask Librarian Bot for paper recommendations by tagging it in a comment: @librarian-bot recommend
Get this paper in your agent:
hf papers read 2611.08175 Don't have the latest CLI?
curl -LsSf https://hf.co/cli/install.sh | bash Models citing this paper 0
No model linking this paper
Datasets citing this paper 0
No dataset linking this paper
Spaces citing this paper 0
No Space linking this paper
Collections including this paper 0
No Collection including this paper