ContractHIL-HLS Automates HLS Design with Hardware Feedback

Jingbo Zhang, Haoxiang Sun, Wenbo Wang, Wenbo Zhang· July 29, 2026 View original

Summary

ContractHIL-HLS is a multi-agent workflow that uses a structured contract and hardware-in-the-loop feedback to automate and improve High-Level Synthesis (HLS) design, translating natural language requirements into robust hardware implementations.

A new multi-agent workflow, ContractHIL-HLS, has been introduced to streamline and enhance High-Level Synthesis (HLS) engineering. This system leverages a structured "contract" as a central artifact, which translates natural language requirements into explicit interfaces, constraints, validation checks, and rollback rules for hardware design. This contract ensures semantic alignment throughout the development process. A key innovation of ContractHIL-HLS is its integration of hardware-in-the-loop feedback. The system incorporates real-world data from HLS tools, Vivado, PYNQ runtime, power consumption, and failure evidence directly back into the design generation process. This extends LLM-assisted HLS beyond just kernel code, enabling system- and board-level closure. The workflow employs specialized agents: a Contract Agent for natural language translation, an HTML Agent for contract rendering, and a Hardware-in-the-Loop Agent for design implementation and revision based on measured evidence. Evaluations show significant improvements in testbench pass rates for small designs and substantial runtime reductions for complex applications like post-quantum cryptography accelerators, demonstrating its practical efficacy.

Why it matters

This system significantly advances automated hardware design, making it faster, more reliable, and more accessible by bridging the gap between natural language requirements and functional hardware with real-world feedback.

How to implement this in your domain

  1. 1Evaluate ContractHIL-HLS or similar contract-aligned multi-agent systems for accelerating hardware design cycles.
  2. 2Explore integrating hardware-in-the-loop feedback mechanisms into existing AI-assisted engineering workflows.
  3. 3Train engineering teams on using structured contracts to define and validate complex system requirements.
  4. 4Pilot AI-driven HLS design for specific, non-critical hardware components to assess efficiency gains.

Who benefits

SemiconductorElectronics ManufacturingAerospace & DefenseAutomotiveAI Hardware

Key takeaways

  • ContractHIL-HLS is a multi-agent workflow for High-Level Synthesis (HLS) design.
  • It uses structured contracts to translate natural language requirements into hardware specifications.
  • Hardware-in-the-loop feedback from real measurements improves design and system-level closure.
  • The system significantly improves design pass rates and reduces runtime for complex hardware.

Original post by Jingbo Zhang, Haoxiang Sun, Wenbo Wang, Wenbo Zhang

"arXiv:2607.25283v1 Announce Type: new Abstract: This paper presents ContractHIL-HLS, a contract-aligned multi-agent workflow for practical high-level synthesis (HLS) engineering. The workflow makes three contributions. First, it introduces a structured contract as the semantic-al…"

View on X

Originally posted by Jingbo Zhang, Haoxiang Sun, Wenbo Wang, Wenbo Zhang on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses