Conformance & attention

Conformance is the deeper half of the compliance lens. Where the light lens reads for simple signals like expiry dates, the conformance engine asks a harder question: for every commitment this project has made, is there evidence in the documents that it was actually met? A commitment with no proof, weak proof, or contradicting proof becomes a gap.

The engine runs as a sweep over the project corpus. It extracts commitments, searches for the evidence that would satisfy each, judges the match, and publishes the results as findings. Every finding keeps the citation it was judged on, so you can always trace a gap back to the exact page.

The engine models two things. A commitment is something the project promised — a control in a safety plan, an inspection required by an ITP, a clause in a specification. A verdictis the engine’s judgment of whether the corpus proves that commitment was met, together with the citations it relied on.

1. 1

   Extract commitments

   Commitments are pulled from intention documents — plans, specifications and ITPs — and from document signals such as required inspections.

2. 2

   Retrieve evidence

   For each commitment, Alloovium searches the project corpus for the records that would satisfy it, filtered to what you have permission to see.

3. 3

   Judge the match

   The engine decides whether the evidence verifies the commitment, and records the reasoning and citations behind that verdict.

4. 4

   Publish findings

   Anything short of verified is published as a gap and routed to a person through the Attention surface.

A gap is any verdict that falls short of fully verified. The engine grades each commitment so you can tell the difference between missing evidence and a genuine contradiction.

Findings carry a severity from low to critical, and roll up into a per-standard health picture — how many commitments are verified, partial, missing or contradicted. That summary is what tells you whether a project is broadly audit-ready or has critical gaps to close first.

Finding a gap is only useful if it reaches the person who can close it. The Attention surface is how gaps become work. When a finding is published, it is attributed to the person tied to the source document and raised as an item they can act on, with a notification so it is not lost.

Attention items appear in the same feed as the rest of your day on Today, tagged as compliance, so you triage them alongside schedule and cost work rather than in a separate silo.

A war-game is a forward-looking check. Instead of grading commitments already made, it asks what would happen if a specific situation arose — and whether the project’s documents show the team is ready for it.

You enter a scenario in plain language, for example bushfire on site or a subcontractor’s insurance lapses mid-works. Alloovium pulls the most relevant document excerpts and asks the model where the response would fall short, returning a short summary and a list of gaps, each with a severity and a recommended action.

1. 1

   Describe the scenario

   Write what could go wrong in plain language.

2. 2

   Alloovium reads the corpus

   It retrieves the most relevant excerpts from the project documents to ground the analysis.

3. 3

   Review the gaps

   You get a summary plus specific gaps. Accept the ones worth tracking and route them to the risk register; dismiss the rest.

Every finding — light signal, conformance gap or accepted war-game gap — moves through the same simple lifecycle so nothing is left half-open.

Once your findings are in good shape, you can package the supporting evidence for an auditor. That is covered on the audit pack and evidence page.