Validate firmware in simulation with AI agents
Hi HN, we're Ahnaf and Seungmin from Simantic (https://simantic.com). Task: Simantic founders, ex-firmware engineers, built a custom Rust simulation engine after finding Renode/QEMU wrappers too slow and inaccurate. The workflow allows AI agents to validate firmware via virtual devboards and CI/CD pipelines without physical hardware. Human engineers define the simulation models and maintain hardware validation rigs to ensure fidelity against real silicon.; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: Simulation currently supports only specific digital MCUs (STM32, ESP32, RP2040) and lacks mixed-signal/analog fidelity. Hardware remains the source of truth for safety-critical systems, meaning simulation cannot fully replace HIL testing. Integration requires adapting existing firmware to the Simantic API and accepting cloud or licensed local execution constraints..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
Hi HN, we're Ahnaf and Seungmin from Simantic (https://simantic.com).
Workflow
Simantic founders, ex-firmware engineers, built a custom Rust simulation engine after finding Renode/QEMU wrappers too slow and inaccurate. The workflow allows AI agents to validate firmware via virtual devboards and CI/CD pipelines without physical hardware. Human engineers define the simulation models and maintain hardware validation rigs to ensure fidelity against real silicon.
Reported result
The source claims validation time dropped from hours to seconds and cites a satellite startup saving 99% cost/time versus hardware dev modules. However, no independent metrics or public silicon validation benchmarks are provided to verify these claims. The authors state they will publish gap reports soon but have not yet done so.
Human role
The original post does not say how a person stayed involved.
Limitations
Simulation currently supports only specific digital MCUs (STM32, ESP32, RP2040) and lacks mixed-signal/analog fidelity. Hardware remains the source of truth for safety-critical systems, meaning simulation cannot fully replace HIL testing. Integration requires adapting existing firmware to the Simantic API and accepting cloud or licensed local execution constraints.
Try it
I am a firmware engineer evaluating Simantic for my [Project Name] using [MCU Model, e.g., STM32F4]. I have a repository at [Repo URL] with existing test suites. Please perform a dry run: 1. Configure the Simantic CI/CD integration for a single branch. 2. Run the existing unit tests against the simulated MCU. 3. Compare the simulation results with a known-good hardware trace if available, or flag any timing discrepancies. 4. Generate a report showing pass/fail status and any detected gaps between simulation and expected behavior. Do not merge to main or deploy to production until I review the report and confirm the simulation fidelity is acceptable for this stage of development.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.