Skip to content

Review AI Actions

Every AI action in DPC Pro is logged (from conversations to patient data lookups to payment recovery notifications) so your team can review, verify, and course-correct.

Transparency is fundamental to trusting AI in a healthcare setting. DPC Pro maintains a detailed audit log of every action the AI assistant performs, including what it did, what data it accessed, and what the outcome was. Managers and clinicians can review this log at any time.

This page covers how to access the AI audit log, what information is recorded for each action, and how to use the log to identify issues or improve AI behavior. The audit log is also valuable for compliance purposes, providing documentation for your practice reports that all AI actions were supervised.

Regular review of AI actions helps your team build confidence in the assistant and catch any behavior that needs adjustment.


AI actions are recorded in the same HIPAA-compliant audit log used for all activity in DPC Pro. To view AI-related entries:

  1. Navigate to Compliance -> Audit Log.
  2. The log displays all recorded actions, sorted by most recent first.
  3. To filter for AI-specific actions, use the Search field and enter AI or conversation.
  4. You can also filter by:
    • Date range: Select a start and end date to narrow results
    • Action type: Filter by Create, Read, Update, Share, or Export
    • User: See actions performed by a specific staff member
    • Patient: See all AI actions that accessed a specific patient’s data
  • Practice managers see the full audit log for their practice, including all AI actions by all staff
  • Clinicians see audit entries related to their own AI conversations and patient data access
  • Organization administrators see audit logs across all practices in the organization


Every AI interaction generates one or more audit log entries. Here is what is recorded for each type of action:

EventAction TypeWhat Is Recorded
New conversation startedCreateUser, practice, AI provider used, whether patient context was included
Practice context shared with AIShareUser, practice name, AI provider (records that practice data was sent to the AI model)
Patient context attachedCreateUser, patient name, note that patient data was included in the AI conversation

When the AI retrieves specific data to answer a question, each retrieval is logged separately:

Data RetrievedAction TypeWhat Is Recorded
Patient detail lookupReadUser, patient name, that the AI accessed the patient’s full record
Patient searchReadUser, search criteria used, number of results returned
Appointment lookupReadUser, date range queried, number of appointments returned
Billing lookupReadUser, whether patient-specific or practice-wide, amount data accessed
Clinical notes lookupReadUser, patient name, number of notes retrieved
EventAction TypeWhat Is Recorded
Document search enabled for conversationUpdateUser, conversation, toggle state
Patient-linked documents shared with external AI providerShareUser, patient name, number of document sources shared, AI provider name
EventAction TypeWhat Is Recorded
Recovery sequence startedCreateMembership, failure reason, invoice reference
Outreach notification sentCreateRecipient email, dunning stage, message template used
Payment recoveredUpdateMembership, recovery timestamp
EventAction TypeWhat Is Recorded
AI settings updatedUpdateUser who made the change, AI provider, settings modified
Guardrails changedUpdateUser, previous value, new value

To examine a specific audit log entry in detail:

  1. Navigate to Compliance -> Audit Log.
  2. Find the entry you want to review (use filters to narrow the list).
  3. Select the entry to open its detail view.

Each audit log entry includes:

  • Timestamp: Exact date and time of the action
  • User email: The staff member who initiated the action
  • User IP address: The network address the action came from
  • Action type: Create, Read, Update, Delete, Share, or Export
  • Action detail: A plain-language description of what happened (for example, “AI tool get_patient_detail accessed patient data” or “AI conversation created with full practice context”)
  • Object: The record that was accessed or modified (conversation, patient, document)
  • Patient: If patient data was involved, the patient’s identity
  • Practice: The practice context the action occurred in
  • Success/failure: Whether the action completed normally or encountered an error

The audit log helps you spot patterns that may indicate a problem with AI usage or configuration.

  • Unusually high patient data access: If a single user is looking up many patients through the AI in a short time, it may indicate misuse or an overly broad question pattern. Review the conversations to understand the context.
  • Frequent errors: Multiple error entries for AI responses suggest a configuration issue (incorrect API key, rate limiting, or model availability). Check Settings -> AI Assistant and run a connection test.
  • Patient data shared with external provider: When using the Claude AI provider, audit entries marked “Share” with patient context indicate that patient information was sent to the external API. Verify this aligns with your practice’s data handling policies.
  • Unexpected users: If a staff member you do not expect to use the AI appears in the log, verify their role and feature access in Settings -> Staff & Roles.

When you identify a concerning pattern:

  1. Review the conversation: Check the AI conversation itself to understand the context of the actions logged
  2. Talk to the user: If a staff member’s usage seems unusual, discuss it with them before changing settings
  3. Adjust guardrails: If the AI is accessing data you would prefer it not access, visit AI Guardrails and Preferences to tighten its configuration
  4. Document your review: Note when you reviewed the audit log and what actions you took. This supports your compliance posture.

You can export audit log entries for compliance documentation, legal requests, or internal review.

  1. Navigate to Compliance -> Audit Log.
  2. Apply any desired filters (date range, action type, user, patient).
  3. Select Export CSV in the top-right corner.
  4. DPC Pro generates a CSV file containing up to 10,000 matching entries.

The exported CSV includes:

ColumnDescription
TimestampDate and time of the action
User EmailStaff member who performed the action
User IPNetwork address
ActionAction type (Create, Read, Update, Share, Export)
Object TypeType of record involved (conversation, patient, document)
ObjectDescription of the specific record
PatientPatient name if PHI was accessed
PracticePractice where the action occurred
Request PathThe system endpoint that processed the action
Request MethodHTTP method (GET, POST)
SuccessWhether the action succeeded
Error MessageError details if the action failed
DetailsAdditional metadata (field changes, parameters)

For HIPAA Right of Access requests or internal compliance reviews, you can export all actions related to a specific patient:

  1. Navigate to Compliance -> Patient Access Report.
  2. Select the patient from the search field.
  3. Choose a date range.
  4. Select Export CSV.

This report shows every time anyone (including the AI) accessed that patient’s data, what type of access occurred, and when.


If you have questions about AI audit logging, reach out to the DPC Pro support team at [email protected] or visit the troubleshooting guide.