Pulse is a real-time dashboard that shows city leaders — mayors, city managers, department heads — what their community is saying, across every engagement channel, without having to open each project individually. It combines participation stats, topic and sentiment tracking, and a geographic view with an AI panel that can read through your comments and write a cited summary of what residents are saying, on request.
Before You Begin
Permissions: Pulse is visible only to users with organization- or department-level access to the account. Project users cannot access Pulse. If your access is scoped to a specific department, you'll only see that department's data; users with full account access see everything.
Setup: None required. Pulse is available to all customers by default and works automatically over your existing projects — no configuration, data mapping, or minimum data threshold.
AI setting: Pulse's AI features (Top Topics and the AI Insights panel) require the account-level "Allow agentic AI" setting to be turned on. This article covers the experience with that setting on. A companion article covers Pulse with it off, and a separate article covers the toggle itself.
How Pulse Works With AI
1. Opening Pulse
Where: Left navigation → Reporting → Pulse, landing on the Dashboard. The rest of Pulse is organized into three more nav sections: Map, Teams, and Data Sources.
What Pulse pulls in automatically: survey responses and open-ended comments, public meeting comments, comment-form submissions, transcribed voicemail messages, and email submissions to your engagement addresses — scoped to your account and, if applicable, your department.
2. The Dashboard
What you see:
Five top-line stats with trend vs. the prior period: Visitors, Participants, Sign-ups, Responses, Active Projects
Top Topics: a quick-read card identifying your top 3–5 community discussion topics right now, each with a short AI-written summary and a sentiment breakdown (positive/neutral/negative)
Engagement Timeline: daily/weekly/monthly activity by channel, toggled across 7-day, 30-day, 90-day, and 1-year views
Top Questions and Top Comments: your most popular poll questions and comments platform-wide
The AI Insights panel, where you ask questions (see below)
What AI does here (Top Topics): reads your qualifying comments for the selected date range (sampling the first 500 and truncating each to 150 characters if there are more, to keep this quick — typically under 5 seconds), identifies 3–5 topics, and for each returns a name, an estimated comment count, a 2–3 sentence summary, and a sentiment split. Results are cached for 10 minutes. Only runs for date ranges of 90 days or less — longer ranges skip it. Nothing to configure; it runs automatically whenever eligible.
3. Asking a question — the AI panel
Where: The AI Insights panel on the Dashboard.
What you do: Type a natural-language question — "What are residents saying about water infrastructure?" — and submit it. No need to pick a project or data source first.
What AI does — the mechanics:
Loads every qualifying comment in scope — no sampling, unlike Top Topics. Hidden, deleted, or empty comments are excluded.
For each comment, reads the text (translated to English automatically if submitted in another language), the project name, the question it responded to, the date, the source channel, and a unique ID for citations.
Processed by an AI model run inside PublicInput's own Azure environment — your comment data doesn't leave that environment or go to a third-party AI vendor. Very large comment volumes are read in a single pass where possible; only for genuinely large accounts does it fall back to a chunked batch-and-combine approach.
The answer streams in live, token-by-token.
Every factual claim is backed by one or more citations — real comment IDs shown inline (e.g., "62% of comments expressed support for expanded bike infrastructure [ID:4521][ID:4533]") — clickable to preview the underlying comment, project, and excerpt.
Alongside the answer: identified themes with sentiment breakdowns, total comments analyzed, date range covered, and a few suggested follow-up questions.
Fully automatic once you ask — no review/approval step before you see the result.
4. Asking a follow-up
Type another question in the same session — "How does sentiment differ between the Northside and Southside districts?" The full conversation so far is sent as context, so follow-ups build on what's already been discussed. Conversations are saved so you can return to one later.
5. Map, Teams, and Data Sources
Map: the same geographic engagement data as a full-screen map, showing where participation is coming from.
Teams: engagement broken down by department/team, so you can compare reach and participation across groups.
Data Sources: a breakdown of which channels are feeding the platform and how much volume each contributes.
None of these three involve AI — they're standard queries either way.
What AI Does
Customer actions: open Pulse, pick a date range/department, ask a question, ask a follow-up.
Automatic (no AI): top-line stats, engagement timeline, top questions/comments lists, map, team breakdown — cached for a few minutes each.
AI actions: Top Topics (sampled, refreshed every 10 minutes); the AI Insights panel (full comment set, cited narrative, streamed, with suggested follow-ups); follow-up Q&A (same, conversational).
Nothing here requires your approval before running or before you see the result.
Reviewing and Managing AI Results
Ask a different or follow-up question at any time — there's no "regenerate" button; asking again is how you refine.
Every citation is clickable, so you can verify any claim against the actual comment behind it.
A limit applies to how many AI questions you can ask in a given period; further requests are blocked until it resets.
Conversations are saved automatically and can be revisited later.
Common Scenarios and Exceptions
Date range longer than 90 days: Top Topics doesn't run; the rest of the dashboard still works.
Non-English comments: automatically translated before analysis.
Very high comment volume: handled behind the scenes; the answer may take longer but isn't cut off.
Asking too many questions in a short window: further questions are blocked until the limit resets.
AI turned off: dashboard, stats, map, and team breakdown remain available — just without Top Topics or the AI Insights panel.
Frequently Asked Questions
Does Pulse replace Mayor Mode?
Yes — Pulse is the successor to Mayor Mode and is intended to fully replace it.
Where does my comment data go when I use the AI panel?
Processed by an AI model run inside PublicInput's own Azure environment — not sent to an external third-party AI service.
Can I trust the AI's numbers?
Every claim is backed by a clickable citation to the actual comment it's based on.
What if I only want the dashboard, not the AI parts?
Turn off "Allow agentic AI" in account settings.
See also: