The Compliance Companion
This article will help you to understand what the compliance companion is, how to use and understand it effectively.
Compliance Companion is a feature that uses AI to evaluate how well your organization's documentation holds against a certification standard.
Instead of manually scouring large document repositories and cross-referencing every clause, Compliance Companion reads your documents and returns a structured assessment for each requirement — telling you what evidence it found, where the gaps are, and how confident it is in its conclusion.
The first supported certifications are:
- ISO 9001:2015 — Quality Management System
- ISO 14001:2015 — Environmental Management System
More certifications will be added over time. Feel free to suggest your needs to your CSM or Support team.
General acess is granted with the Compliance Intelligence Solver and the ''Enable AI Features'' must be checked in Application Settings (under management).
Module permissions > Compliance companion is also used to ensure the right people have the correct access.
In This Article:
- Getting Started
- Navigating the Compliance Companion
- Running an Evaluation
- Getting the most out of the Compliance Companion
Getting Started
You can find the compliance companion under menu options Documents > Comlpiance Companion
The main Compliance Companion page shows every supported certification as a row in a table. From here you can see your latest results, run new evaluations, and delete old ones.

Each row represents one certification standard. The table shows:
- Certification — the name and version of the certification.
- Last run — the date and time the most recent evaluation was completed. Empty if the certification has never been evaluated.
- Score — the overall compliance score from the most recent evaluation, shown as a percentage with a progress bar. Empty if never evaluated.
- Actions — buttons to run or delete an evaluation for that row
Navigating the Compliance Companion
💡Click “Run evaluation” (arrow icon) in any row to start an evaluation for that certification.
💡Click on any row that has a completed evaluation to open the evaluation details page. Rows with no evaluation are not clickable.
💡To remove an evaluation result, click the delete icon (trash icon) in the Actions column of the relevant row. A confirmation dialog will ask you to confirm before anything is deleted.
💡Deleting an evaluation clears the results for that certification but does not remove it from the table. The row remains, and you can run a fresh evaluation at any time.
💡Certifications with a completed evaluation appear at the top of the table, sorted from most recent to least. Certifications that have never been evaluated appear at the bottom.
Running an Evaluation
Click the “Run evaluation” (arrow icon) in any row to start an evaluation for that certification.
A. If this is the first evaluation for that certification :
A confirmation dialog appears. It will tell you which certification you’re about to evaluate and that the process may take a few minutes. Click “Run evaluation” to confirm or “Cancel” to go back.
B. If an evaluation already exists:
As only one evaluation per certification is stored at a time, running a new one will replace the previous result. A warning dialog appears showing the certification name and the date of the existing evaluation. Click “Overwrite” to proceed or “Cancel” to keep the existing result.
Overwriting an evaluation is permanent. The previous results cannot be recovered once replaced.
Once confirmed, the evaluation runs in the background. The table where the evaluation clauses are going to be shown displays a spinner while the clauses are being assessed.

Evaluations can take several minutes depending on the size of your document store and the number of clauses.
If something goes wrong during the evaluation, no results are saved. An error message will tell you what happened. Any previous evaluation for that certification is preserved.
Reading an Evaluation
The evaluation detail page gives you a full picture of how your documentation performs against a certification standard.

At the top of the page, you’ll find a summary card showing:
- Overall compliance score — a percentage and progress bar representing the combined result across all clauses.
- Compliant count — the number of clauses fully satisfied by your documentation.
- Partially compliant count — the number of clauses where some evidence was found, but gaps remain.
- Not compliant count — the number of clauses where the documentation did not satisfy the requirement.
How the Score is calculated
Each clause contributes to the total score. The contribution depends on its compliance status:
|
Compliant |
Full points |
|
Partially compliant |
Half points |
|
Not compliant |
No points |
For example:
An evaluation with 10 clauses where
6 are Compliant = 6 × 10 (full points)
2 are Partially compliant = 2 × 5 (half points)
2 are Not compliant = 2 × 0 (no points)
would score 70%
(6 × 10) + (2 × 5) + (2 × 0) = 70
Colour Coding
Badges throughout the evaluation use a consistent colour system, Green, Yellow and Red.


AI-flagged compliance results with confidence levels help you focus review time where it matters, trusting high-confidence findings while catching gaps that need a human look.
The Clause List
Below the summary, every clause in the certification is listed in order. Each row shows the clause number, category, a short description of what the clause requires, and its compliance status and confidence badges.

Use the filter controls above the list to show only clauses with a particular compliance status. You can select one or more statuses at the same time. To return to the full list, clear the filter.
If no clauses match the selected filter, the list will show an empty state. Try a different filter or clear the filter to see all the clauses.
The Clause Detail Panel
Click on any clause row to open a panel on the right-hand side of the screen showing the full assessment for that clause.

The panel includes:
- The compliance status and confidence level, prominently displayed at the top.
- Audit check — the full text of the requirement being assessed.
- Evidence used — the specific documents or excerpts the AI found.
- Reasoning — the AI’s explanation of its conclusion.
- Gaps or uncertainties — what is missing or unclear, if anything.
Close the panel by clicking the X button in the panel header, or by clicking anywhere on the dimmed area behind the panel.
Re-running an Evaluation
The ''Run again'' button at the top of the evaluation detail page lets you refresh the evaluation after you have updated your documentation.

Because a result already exists, you will always see the overwrite confirmation dialog before the new evaluation starts

Getting the Most out of the Compliance Companion
1. Document Coverage Matters
Compliance Companion can only assess what it can find. If relevant documents are missing from your AlisQI document store, or are stored in a way that makes them hard to retrieve, the AI may return Not compliant or Low confidence results even when your organisation does meet the requirement.
- Make sure key policy documents, procedures, and records are uploaded to AlisQI before running an evaluation.
- Use descriptive document titles and keep documents up to date — the AI uses document content, not just filenames.
- If a clause comes back as Not compliant with Low confidence and you believe you have the relevant documentation, check that the document is in your store and try re-running the evaluation.
2. Interpreting partial and low-confidence results
Partially compliant and Low confidence results are the most actionable outputs of an evaluation. They point to specific clauses where documentation exists but is incomplete, ambiguous, or inconsistently worded.
- Use the Gaps or uncertainties section in the clause detail panel to understand exactly what the AI found lacking.
- Use the Evidence used section to verify whether the AI found the right documents, or whether it retrieved something unrelated.
- Low confidence on a Compliant result is a signal to review the evidence manually — the AI may have made a weak inference.
3. When to re-run?
Re-running an evaluation permanently overwrites the previous result. Run again when:
- You have added or updated documents that address identified gaps.
- A significant amount of time has passed and you want a fresh picture.
Avoid re-running speculatively — if your documentation has not changed, the result is unlikely to be materially different.