Appearance
Control Assessment Agent
The Control Assessment Agent makes a readiness assessment of a Control. It reads the Control definition, the processed Evidence linked to the Control, and the published policies linked to the Control.
What this is
When you run the agent, it gives:
- a readiness score from 0 to 100
- a short summary of what the current Evidence shows and what it does not show
- gaps, each with an analysis and references
- recommendations, each with references
- the References that it used: the Evidence items and policies that it read
Modulos stores the assessment when the run completes. There is no save step. The Assessment tab shows the latest assessment. Each run also adds a line with its readiness score to the Comments and Logs tab of the Control. The agent does not change the status of a Control. You stay accountable for how you use the assessment: compare it with the Evidence, and run the agent again after you correct the gaps.
Where in Modulos
You run the Control Assessment Agent from the page of a Control, or for all Controls of a framework as a Job. The dashboard and the Controls page show which Controls have an assessment.
AI System → Controls → select a Control → Assessment → AI Agentruns the agent.AI System → Controls → Create Jobassesses the Controls of the frameworks that you select, by hand or on a schedule.AI System → Dashboardshows the AI Assessment column of the summary card: how many Controls have an assessment, and their average readiness.AI System → Controls, filter bar: the AI Assessment filter shows the Controls that are Assessed or Not assessed.

The Assessment tab shows the latest assessment: a readiness score and summary, the References, the gaps, and the recommendations.
- AI Agent Runs the agent again. The new assessment becomes the latest one.
- Control Readiness The readiness score and a summary of what the Evidence supports.
- References The Evidence items and Policies that the agent read.
Who can do what
Permissions
- AI System Owners and Editors can run the agent and create assessment Jobs.
- All members of the AI System, Viewers too, can read the stored assessments and follow the references to the items.
The AI Agent button shows only while the Control is Not Executed. If you do not see the AI Agent button or the Assessment tab, speak to your AI System Owner or organization admin.
How it works
When you run the agent, Modulos does these steps:
- It takes the question, description, and guidance of the Control.
- It collects the processed Evidence items linked to the Control. It also collects all published policies linked to the Control in Associated Policies.
- It reads all of that material for the assessment. If the material is too long for one assessment, the agent reads only the first part of the longest documents.
- It writes a structured assessment in the language of the AI System.
The references in gaps and recommendations link to the Evidence item or the policy, with a preview when you hover. A deleted item, or an item that you cannot open, shows without a link.
Processing gate
Modulos processes uploaded Evidence and policy files before the agent can read them. See Evidence. The agent reads only the linked Evidence items that Modulos processed. When it leaves out other linked items, the Assessment tab shows the notice n evidence documents were not used in this assessment, with each item and its reason:
- still being processed: Modulos did not complete the processing yet.
- waiting for AI quota: Modulos processes the item automatically when the AI quota of the organization has space again.
- could not be processed: the item shows Processing failed. Remove it from the Control. If necessary, upload a copy of the file that Modulos can read.
- not indexed yet, or not processed while AI features were off.
Because of processing, Modulos refuses a manual run in two conditions:
- The Control has Evidence, but Modulos processed none of it. The message is "None of this control's evidence is ready to be used in the assessment". If the items are still in processing, wait and run the agent again. If they show Processing failed, waiting does not help: remove them from the Control, or upload copies that Modulos can read.
- A linked policy file is still in processing. The message is "A policy linked to this control is still being processed. Please try again shortly." Wait and run the agent again.
A Control with no Evidence is not refused: the agent reports the gaps.
In a Job, each step assesses the processed Evidence of its Control in the same way. A step fails with the message "The control's evidence is still being processed" only in the two conditions above, and the next run of the Job tries again.
When the AI quota of the organization is used up, a manual run shows the error "Something went wrong.", and a Job step fails with "The organization's AI usage quota did not allow this control assessment". See AI usage.
Framework-wide assessment Jobs
AI System → Controls → Create Job opens Create Control AI Assessment Job, where you select frameworks. The Job stores the frameworks, not a fixed set of Controls. Each run of the Job reads the Controls of these frameworks again and makes one assessment step for each Control that is Not Executed. The run leaves out Executed, Out of Scope, and Archived Controls. So a Control that you add to a framework after you create the Job becomes part of the next run.
Each step stores a new assessment for its Control. The Job overview shows the Controls of each framework with their code and their Control Readiness. Each step of a run shows its readiness score. For schedules and runs, see Jobs.
Tracking coverage
The AI Assessment column of the AI System dashboard shows how many Controls have an assessment and their average readiness. It shows when the agent is available in your organization, and it uses the latest assessment of each Control. See AI System Dashboard.

The AI Assessment column of the AI System dashboard shows how many Controls have an assessment and their average readiness.
- AI Assessment The Controls with a stored assessment, out of all Controls that count.
- Average readiness The average readiness score of the assessed Controls.
- View Assessed Controls Opens the assessed Controls, the lowest readiness score first.
How to use it
- Attach the most applicable Evidence to the Control, and link the policies for the Control.
- Open the Assessment tab of the Control and select AI Agent. Modulos stores the result when the run is complete.
- If Modulos refuses the run because files are still in processing, wait and run the agent again. For Evidence with Processing failed, or Evidence that the agent did not use, see Processing gate.
- Read the summary. Then compare the gaps and recommendations with the Evidence in the references.
- Update the Evidence and the Control report as necessary, and then run the agent again.
To assess many Controls, use AI System → Controls → Create Job and select frameworks. To find the Controls that have no assessment, use the AI Assessment filter on the Controls page and select Not assessed. When the AI System has a lifecycle stage, the Controls page opens at it: remove the Lifecycle Stage filter to include all lifecycle stages.
Important considerations
- The readiness score is not a certification. Use it to find the Controls that have the most gaps. Do not use it as an approval decision.
- If Evidence is missing, the agent reports gaps. Use the gaps as a checklist to collect the correct items.
- Run the assessment before you set a Control to Executed. The AI Agent button shows only while the Control is Not Executed, and the Evidence of an Executed Control is locked.
- AI can make mistakes. Compare the claims and references with the Evidence.