Evaluate work items with LLM — rich graph actions from standard clauses

After drafting a hazard, safety goal, and a few requirements, the natural question is what comes next: which artifacts are missing, which clauses of the standard apply, and what should be created or revised to keep the traceability chain intact.

WCGW Safety Suite's Evaluate work item with LLM feature answers those questions — and lets you act on them.

How to?

Select any work item that you have created and run an LLM evaluation by clicking "Evaluate" under the "Evaluate work item with LLM" section. The model analyses the item against the relevant functional safety standards — ISO 26262, IEC 61508, ISO 13849 — and produces a list of suggested actions.

Each suggestion includes:

• The action: create, revise, or link
• The payload: the actual work items — complete with IDs, ASIL levels, phases, and standards
• The reasoning: a direct quote of the clause (e.g. "ISO 26262-5 Clause 6 requires the specification of hardware safety requirements derived from the technical safety concept")

An example in practice

Evaluating a set of technical safety requirements for a valve driver, the LLM produced three high-importance suggestions:

1. Create hardware safety requirements — ISO 26262-5 Clause 6 Three new HWSRs for the Smart MOSFET output stage, absolute encoder interface, and pressure transducer analog interface, each with full draft markdown covering thresholds, diagnostic feedback, and transient protection.

2. Create software test cases — ISO 26262-6 Clause 10 Two test cases for control loop timing and interlock state machine verification, with detailed test procedures and pass criteria.

3. Revise Safety Goal SG-03 and FSR-03 — ISO 13849-1 The LLM detected that the stabiliser interlock needed explicit PL D / Category 3 specification and provided an update for both the safety goal and the functional requirement with the correct architecture and diagnostic coverage targets.

All three actions are powered by ready-to-use payloads that can be incorporated in the relevant work item.

Standards-backed graph updates that you sign or discard

The LLM produces rich graph updates — collections of concepts and relations that slot directly into the traceability graph. Each action is backed by a specific clause. Every proposed create, revise, or link operation carries the full payload needed to integrate into the project's traceability model. But to maintain the audit trail, each update must be signed by the user clicking "Act". This guarantees that every change in the graph is attributable and non-repudiable — exactly what a functional safety process demands.

This turns an open-ended "what's missing?" moment into a concrete, standards-aligned action list in seconds — one that preserves accountability at every step.

Want to see what it finds in your project? Contact us for a demo.