Skip to main content

Inbox AI Moderation

Automatically classify and act on comments and DMs with AI labels, rules, and audit trails.

What Inbox AI Moderation does

AI moderation automatically classifies incoming comments and DMs into categories (spam, promotional, offensive, etc.) and executes rules you set (hide, flag, escalate). Every action is audited and reversible.

Important: Moderation is opt-in and default OFF. Nothing is hidden, flagged, or escalated until you enable it.

Before you start

  • AI moderation requires a paid plan
  • Each classification costs AI credits (default budget: 200 per day)
  • When the daily budget is exceeded, classification falls back to heuristics (free, but less accurate)
  • You control which rules are active per category

Step 1 — Enable AI moderation

  1. Go to Inbox → Moderation tab (or Inbox → gear icon → Moderation settings)
  2. Toggle Enable AI moderation ON
  3. Set your daily AI budget (default: 200 uses per day)
  4. Click Save

Note: Until you create and enable a moderation rule, nothing happens — all items remain visible by default.

Step 2 — Set up moderation rules

Rules live in the Inbox Automations tab under Moderation:

  1. Click + New Moderation Rule
  2. Choose your category:
    • Approval requests — identify comments/DMs asking to approve posts
    • Publish results — notifications about post outcomes
    • Warnings — potentially harmful or policy-violating content
  3. Pick the classification label (e.g., "Spam", "Offensive", "Promotional")
  4. Set minimum confidence (how confident the AI must be to trigger the rule: 50%, 75%, 90%)
  5. Choose the action:
    • Hide — remove from your inbox view (only works on Instagram/Facebook comments; other platforms downgrade to Flag)
    • Flag — mark for your review with a label
    • Escalate — send an alert (future: desktop/email notifications)
  6. Click Save

Tip: Start with Flag until you're confident in the rules, then upgrade to Hide for automated cleanup.

Step 3 — Monitor classified items

When items are classified:

  1. A moderation chip appears on each item showing the label (e.g., "Spam - 92% confidence")
  2. Filter your inbox by classification using the Moderation filter (appears only when classified items exist)
  3. Review what the AI found before hidden items disappear
  4. See the action history at the bottom (which rules executed and when)

Step 4 — Restore items (undo automated actions)

If a moderation action was wrong:

  1. Open the item's detail view
  2. Click Restore (if hidden) or Undo (if flagged)
  3. The item reappears in your inbox
  4. The restore is logged as a manual action in your audit trail

Note: Restore only works for items hidden by Aidelly; platform-hidden items cannot be unhidden from Aidelly.

Classification labels (what the AI looks for)

The AI classifies comments and DMs into these categories:

  • Spam — low-effort, repetitive, or bot-like messages
  • Promotional — self-promotional or overly salesy content
  • Offensive — potentially rude, disrespectful, or harmful language
  • Informational — neutral questions or informational replies
  • Positive — genuinely supportive or complimentary messages

Your rules match against these labels.

Platform support matrix

The Hide action only works on certain platforms:

PlatformSupport
Instagram (comments)Full support
Facebook (comments)Full support
Facebook (messages)Not supported; degrades to Flag
Other platformsNot supported; degrades to Flag

When Hide degrades to Flag on unsupported platforms, the audit trail notes the degradation.

Daily AI budget

  • Default: 200 classifications per day
  • Configurable: Adjust in Moderation settings
  • When exceeded: Classification switches to free heuristics (less accurate, no AI cost)
  • Resets daily at midnight in your timezone

Tip: Monitor your daily usage in Account Settings → Usage to stay within budget.

Audit trail and reversibility

Every moderation action is recorded in Workspace Activity:

  • Automated actions: Actor = "Automation" (which rule ran)
  • Manual actions: Actor = your name (when you restore or flag manually)
  • Timestamp: Exact time the action occurred
  • Reversible: You can undo any automated or manual action

Access the audit trail via:

  • Inbox item detail panel → Action history section
  • Account → Activity Log → filter by "Moderation"

Common questions

Q: Will moderation affect my relationships with followers?
A: Hiding comments only removes them from your inbox view. On Instagram/Facebook, the original commenter still sees their own comment. Flag and Escalate don't hide anything.

Q: What if the AI misclassifies something?
A: Click Restore to undo. The item reappears and you can review it manually.

Q: Can I turn off moderation for specific platforms?
A: Not yet. Moderation rules apply to all platforms where applicable. If you want to disable for one platform, set a very high confidence threshold (90%+) or delete the rule.

Q: What counts toward the daily budget?
A: Each new incoming item costs 1 use when classified. Re-reviewing an already-classified item doesn't cost extra.

Q: Can I set rules per platform or per account?
A: Not yet. Rules apply workspace-wide. Platform-specific rules are a future improvement.

What to do next