TraceBoard's engineering data — requirements, links, test results, baselines — is deterministic and customer-controlled. AI is an optional layer on top of that core: it can propose, draft, and flag, but nothing enters your system of record without your explicit approval. The deterministic core can operate without a configured AI model; fully local operation depends on the deployment configuration and any external integrations selected.
Most ALM vendors handle AI one of two ways, and neither works for regulated teams.
Legacy tools add a chat box or an "AI assist" button as a roadmap checkbox, with no architectural line between what the model suggests and what the system treats as fact. It looks modern in a demo and becomes a liability in an audit.
Newer tools build the model into the core workflow itself — convenient, until a compliance engineer has to explain to an auditor why a traceability link exists because "the model said so." For a system of record, that's not an acceptable answer.
Requirements, links, and test data are structured facts placed there by a person — never model output treated as ground truth.
Nothing an AI feature proposes becomes part of your record until a human reviews and accepts it. No silent writes, ever.
From a single suggestion to the entire deployment, AI can be switched off with no loss of core functionality — not a degraded mode, the same product.
The core suite can be self-hosted, and a local model can keep AI requests inside your network when configured that way. OIDC, remote Git and cloud LLMs require their external services.
This four-stage pipeline illustrates how TraceBoard handles AI-assisted workflows, including test case generation. The exact interaction depends on the feature.
Given a requirement's text and context, the model drafts candidate test cases — nothing is written to the project yet.
Drafts are checked against your schema — required fields, valid states, correct linkage — before a person ever sees them.
Every draft is shown clearly as a suggestion. Accept it, edit it, or discard it — the choice is always visible and always yours.
Once approved, the test case becomes a normal TraceBoard object with the review decision recorded. The committed record is deterministic; approval does not guarantee that the underlying engineering judgment is correct.
Every item below only proposes. Nothing here writes to your project without approval.
Surfaces coverage gaps across requirements, tests, and documents, and helps you prioritize which ones actually matter.
Bring in inconsistent spreadsheets or documents as they are. AI proposes a clean, mapped structure — you approve it before anything imports.
Drafts test cases from requirements and flags existing tests that may be stale after a requirement changes.
Drafts report sections from your project data and flags inconsistencies across your document set.
Structured, typed data — never generated on the fly by a model.
Every change traces back to a specific human decision, including approved AI suggestions.
Coverage, sign-offs, and traceability reports reflect only what was approved — nothing pending, nothing inferred.
Stored items, relationships, calculations, baselines and recorded evidence do not require an AI model to remain inspectable.
TraceBoard deploys self-hosted. A local AI endpoint can support local operation, while external identity, remote Git and cloud LLM integrations require network access.
Configure AI through a supported local or OpenAI-compatible endpoint. Keep engineering data on customer-controlled infrastructure and use the data operations implemented for each workflow.
The product supports local Ollama or OpenAI-compatible HTTP configuration. Generated output depends on the selected model, configuration, context and review.
Smaller models (≤7B) may produce less detailed narratives, while larger models provide deeper insights — giving you full control over the trade-off between cost, latency, and quality.
Use ReqIF 1.0/1.2 import with preview and mapping, plus export for live project or baseline scope. Other supported endpoint-specific operations include JSON, CSV, XLSX, PDF, DOCX and Markdown.
The best way to understand the approval gate is to watch it work on your own data — a messy spreadsheet, a real requirement set, or a document you're tired of formatting by hand.