TRASE

The Foundation for Autonomous Engineering

Forge, Einride, DREV and Chalmers Industriteknik are collaborating on TRASE, supported by 4.5 MSEK in funding from Vinnova, Sweden's innovation agency. The project develops and deploys agentic systems for safety-critical R&D.

Agentic AI is making the generation of engineering artifacts faster and cheaper. Requirements, analyses, code, and test cases can now be produced at a scale that was previously impossible. As generation accelerates, verification becomes the constraint. In safety-critical R&D, every artifact has to be aligned against requirements, applicable standards, system architecture, design decisions, test evidence, and operational feedback. The relationships among these sources form the engineering context needed to verify agent output. Forge treats this context as the verification layer for agentic engineering.

Verification feedback and operational Validation reveal gaps, informing requirements and driving new system design, implementation, and tests. Once reviewed and approved, changes enter verification & validation and eventually operation, generating new evidence and engineering feedback. This creates an engineering flywheel in which every cycle strengthens the context available for the next, speeding up development. Agents help teams complete these cycles faster, while verification loops and review gates keep output aligned with engineering intent. TRASE applies this approach across the engineering lifecycle at Einride and DREV, with Forge Studio connecting engineering data across tools and teams, giving agents the context needed to reason across and contribute to the entire development lifecycle.

In addition to R&D deployments, Chalmers Industriteknik leads the development of a shared benchmarking framework for evaluating these capabilities. Existing benchmarks measure general reasoning or code generation, not engineering, where output must hold up against physical constraints, safety analyses, regulatory approval, and the realities of manufacturing and operation. Reliable data on agent engineering capabilities is difficult to obtain, yet teams need to know what agents can actually do, in which workflows, and in which engineering domains. A shared benchmark provides that evidence as a common reference for the industry.

TRASE combines industrial deployment, systematic evaluation, and accountable oversight to establish a rigorous foundation for autonomous engineering. Forge provides the engineering infrastructure to help teams develop and deploy safe and reliable systems faster, respond with agility to change and preserve human authority over critical decisions.