Automations
Configuring Herald Shield
Herald Shield is the AI-driven enforcement automation. Configure your escalation ladder — what action happens at each offense level, and whether the system fines automatically or flags for board review.
Last updated April 29, 2026
Herald Shield is the AI agent that handles repeat violations. The conceptual overview is in Escalations and repeat offenders. This article is the configuration playbook.
What you’re configuring
Herald Shield has two main pieces:
- Auto-dismiss no-violation cases — when AI analysis concludes a complaint isn’t actually a violation, automatically dismiss it
- Escalation ladder — for confirmed violations, the sequence of actions that fire as offenses repeat
You configure both before turning Herald Shield on.
Step 1 — Enable Herald Shield
Configuration → Herald Shield Automation
Toggle the automation on.
The setting Auto-dismiss no-violation cases controls whether AI-classified-as-no-violation complaints are automatically dismissed. If on, the AI dismisses; if off, the AI’s analysis is recorded but a board member still reviews.
For new users, leave auto-dismiss off at first. Watch what the AI is recommending for a few weeks before letting it act on its own.
Step 2 — Configure the escalation ladder
The ladder is a sequence of escalation steps. Each step represents an offense level. When a violation is confirmed against a property, Herald Shield uses the ladder to determine what action to take based on how many prior offenses that property has had.
For each step:
| Field | Purpose |
|---|---|
| Offense number | 1 = first offense, 2 = second offense, etc. The last step applies to all subsequent offenses. |
| Action | Warning / Formal Warning / Fine |
| Channels | How to notify (email, in-app, push, letter) |
| Fine amount | Only for Fine actions — dollar amount |
| Use covenant fine | Pull the fine amount from the matched covenant rule instead of a fixed amount |
| Fine mode | Auto = fine and close automatically; Flag for review = create the violation and let a board member decide |
| Label | Display name (e.g., “1st Offense — Warning”) |
A typical configuration:
| Offense | Action | Notes |
|---|---|---|
| 1 | Warning | Friendly notice via email |
| 2 | Formal Warning | More severe; certified letter via Physical Mail |
| 3 | Fine, $50, Flag for review | Board approves before fine applies |
| 4+ | Fine, $100, Auto | Repeat offenders get auto-fined |
The pattern: low-stakes tiers (notices, warnings) fire automatically; fines either flag for review or fire on auto, depending on how confident you are in the rules.
Fine mode — auto vs. flag for review
For Fine actions, you have two modes:
Auto — Herald Shield fines the property automatically. The fine appears on the ledger; the resident is notified.
Flag for review — Herald Shield identifies the violation and proposes a fine, but a board member decides whether to actually issue it. The violation appears in the admin queue with the proposed action.
Recommendation: start with Flag for review on all Fine-tier steps. Once you’ve seen a few cycles and trust the AI’s classifications, switch the routine tiers to Auto.
Step 3 — Channels
Each step has channels for how the resident is notified:
- Email — reaches everyone with an email on file
- In-app — visible in their portal
- Letter (Physical Mail) — physical mail; useful for severe tiers requiring formal notice
Use email + in-app for the early tiers; add letter for serious escalations.
What Herald Shield does in practice
When a complaint is filed against a property:
- AI analyzes the complaint against your covenant rules
- If no violation is detected and auto-dismiss is on, complaint is dismissed
- If a violation is detected, Herald Shield checks the property’s prior violation history
- The matching escalation step is identified based on offense count
- The configured action fires (warning sent, fine applied, or flagged for review)
- If Flag for review, a board member completes the action manually
Watching what Herald Shield does
Herald Shield logs each action it takes — including the AI’s reasoning, the matched covenant rule, and the action taken or recommended. Review these regularly:
- Catch wrong classifications before they propagate
- Verify your escalation tiers are firing correctly
- Spot opportunities to refine your covenant rules
Spend 10-15 minutes a week reviewing Herald Shield’s recent activity for the first month or two.
Skipping Herald Shield on specific cases
For violations where the AI shouldn’t escalate — active negotiation, hardship cases, disputed bylaw interpretation — handle the violation manually outside the Herald Shield flow. The board can resolve, dismiss, or fine without Herald Shield’s involvement.
Common mistakes
”Configured the first offense as a fine”
Means the first violation gets fined immediately. Bad — bypasses warnings, residents will dispute. Offense 1 should always be a warning.
”Set fine mode to Auto without testing”
You haven’t seen what Herald Shield is actually deciding. Stay in Flag for review for fine-level actions until you trust the AI’s accuracy on real cases.
”Herald Shield is too aggressive”
Adjust the escalation ladder — move fine actions later in the sequence (so more warnings fire first), or switch them to Flag for review.
”Herald Shield is too lenient”
Inverse — move fine actions earlier, increase fine amounts at later tiers, or switch from Flag for review to Auto for routine cases.
Where to go next
- Escalations and repeat offenders — the conceptual overview
- Issuing a fine — the manual flow Herald Shield is automating
- Dues Chaser — similar automation model for dues collection