Herald Chat

Herald Chat is the AI assistant residents can chat with from their portal. It answers questions about community rules, dues, amenities, and procedures — drawing on documents you’ve uploaded and announcements you’ve made. When configured well, it cuts board email volume by 30-50%.

What it can do

Herald Chat can answer questions like:

• “What are the pool hours?”
• “How do I pay my dues?”
• “Can I have chickens?”
• “When is the next board meeting?”
• “What’s the late fee policy?”
• “How do I book the clubhouse?”

Source material: your community’s covenants, rules, recent announcements, and any external documents you’ve added to the AI knowledge base.

What it can’t (and shouldn’t) do

By design, Herald Chat does not:

• File requests on behalf of residents (only suggests they file one)
• Make decisions about specific situations (“Can I do X?” — it’ll say “Per the rules… but please confirm with the board”)
• Access financial data for a specific resident (privacy)
• Replace the board for legal questions
• Create binding commitments

This is intentional. AI mistakes on specific advice are bad; AI mistakes on general info are correctable.

Setup

Configuration → Herald Chat

Toggle on. The basic version works immediately with built-in HOA knowledge. To make it actually useful for your community, you need to feed it your documents.

Feeding Herald Chat

The AI knowledge base (dedicated article) is what Herald Chat draws from. The setup:

1. Upload your covenants — the CC&Rs PDF
2. Upload your bylaws — the bylaws PDF
3. Upload community rules — pool rules, parking rules, architectural guidelines
4. Index your announcements — the system can include recent announcements automatically
5. Optionally crawl an external website — if your HOA has a public site with FAQs, point Herald Chat at it

Herald Chat reads everything you’ve uploaded and uses it to answer questions. Better source material = better answers.

Where residents access it

Two places:

• Portal home page — chat widget visible to logged-in residents
• In-app help button — same widget, accessible from any page

Residents start a chat, type their question, get an answer in real-time. The conversation is saved in their account history.

What residents see

When a resident asks a question, Herald Chat:

1. Searches the knowledge base for relevant info
2. Composes an answer in plain language
3. Cites the source (“Per CC&R Section 4.3…”)
4. Suggests follow-up actions if relevant (“If you’d like to dispute this, file a complaint here”)

Residents see a normal chat interface. No “I’m an AI” banner unless they ask — but the chat does mention it’s an automated assistant in its first message.

Configuring tone and behavior

Configuration → Herald Chat → Settings

You can tune:

• Tone — Formal / Casual / Friendly. “Friendly” works for most communities.
• Response length — Short (1-2 sentences) / Medium / Detailed. Medium is the default.
• When unsure, response — What to say when the AI can’t find an answer. Default: “I’m not sure about that. The board can help — would you like me to file a question for you?”
• Disclaimer — Optional disclaimer to include with responses on legal/financial topics. Recommended.

A good default disclaimer:

“This is an automated assistant. For specific situations or formal decisions, please contact the board directly.”

Routing escalations

When Herald Chat can’t answer a question (genuinely doesn’t know, or the resident pushes back), it offers to escalate:

“I’m not sure I have the right answer for this. Would you like me to send your question to the board for a follow-up?”

If the resident says yes:

• Herald Chat creates a General-type request (see The four request types)
• The conversation is attached as the request body
• The board sees it in the regular request queue

This means residents don’t have to leave the chat to file a question — Herald Chat does it for them.

When Herald Chat says something wrong

AI mistakes happen. When you spot a wrong answer:

• Update the AI knowledge base sources — make sure the source content is accurate
• Reach out to the resident directly with the correct info
• If the wrong answer caused real harm (e.g., resident was told they could do X and acted on it), apologize directly and make it right

The best defense against wrong AI answers is keeping the knowledge base sources accurate. The AI uses your sources as truth — if your sources are wrong, the AI will be too.

Plan implications

Herald Chat is included on Automate plan and above. Free plan doesn’t include it.

Conversation count limits apply per plan tier. Most communities don’t hit limits unless they’re very active.

Common situations

”Herald Chat told a resident the wrong pool rules”

Two things:

1. Update the source document (your pool rules PDF) with clearer language
2. Mark the conversation as wrong with the correct answer
3. Reach out to the resident directly with the right info

The AI uses your docs as truth. If your docs are wrong, the AI will be too.

”Residents are asking Herald Chat the same questions over and over”

Means it’s working — it’s catching questions that would otherwise be email. You can review the patterns under Conversations and use them to:

• Update your knowledge base for clarity
• Write a community-wide announcement covering the FAQ
• Add an FAQ section to your community website

”Herald Chat keeps escalating questions to the board that we don’t want to handle”

Two fixes:

1. Improve the knowledge base — add clearer documents on the topics being escalated
2. Adjust the “When unsure” template — make it clearer what topics the board will and won’t take questions on

”We don’t trust AI for our community — turn it off”

That’s fine. Disable Herald Chat under Settings. Residents lose the chat option but can still email the board. No data lost.

Where to go next

• AI knowledge base — the source material Herald Chat uses
• The unified email inbox — what Herald Chat is offloading
• The four request types — what gets created on escalation
• Herald Chat feature page (marketing): /features/herald-chat/