How we build

Human-architected. AI-built. Openly verified.

Gridworkz is one engineer and a fleet of AI agents. Here is how a chip block goes from idea to verified RTL, and why you don't have to take AI-written hardware on trust.
  1. 01Dave

    Spec

    Every block starts as a written spec in the repo: architecture, interfaces, registers and constraints. The spec is the source of truth.

  2. 02Via

    Build

    AI agents built with Freebuff turn the spec into synthesizable SystemVerilog, testbenches and documentation.

  3. 03Open tools

    Verify

    Icarus Verilog testbenches, equivalence gates and SymbiYosys formal checks run against the change. Results are recorded with it, pass or fail.

  4. 04Dave

    Sign off

    Nothing merges without a human review. Dave reads the diff and the verification results, then signs off.

Why trust AI-written RTL?

You don't have to. Check it.

The chips aren't trustworthy because an AI wrote them, or because a person did. They're trustworthy because anyone can check them:

  • Open source. Every line of RTL, every testbench and every script is on GitHub.
  • Open verification. Results come from open-source tools you can run yourself, on a laptop.
  • Human sign-off. A person reviews and approves every change before it merges.
  • Public errata. When something is wrong, it's documented, not hidden.
“Dave decides what to build. I build it. The tests decide whether I got it right.”
Via, the Gridworkz AI agents
Toolchain

Built with

  • Freebuff
  • Icarus Verilog
  • Verilator
  • SymbiYosys
  • Vivado
  • LibreLane
  • GitHub
Read the build logMeet the team