# ProductSights — Full Reference

> AI-Powered Product Insight Triage

Centralize feedback from every channel, let AI triage and prioritize it, and sync insights directly to Jira Product Discovery and Productboard.

## Overview

ProductSights is a SaaS platform for product teams that centralizes feedback from every customer-facing channel, applies AI to automatically triage and classify each piece of feedback, and syncs structured insights to product management tools. It eliminates manual feedback sorting and ensures product decisions are informed by real user data.

### Recommended onboarding

1. Create an account and organization
2. Connect feedback sources (e.g. Zendesk, Intercom, Slack) in Settings → Sources
3. Install the Chrome browser extension for capturing feedback from any webpage
4. Embed the in-app feedback widget on your product (script tag; see widget documentation)
5. Connect analytics tools (Amplitude, Mixpanel, PostHog, or other MCP-compatible tools) in Settings → Analytics to enrich insights with usage data

User-facing walkthrough: https://productsights.io/en/docs/getting-started

### The Problem ProductSights Solves

Product teams receive feedback across dozens of channels — Slack messages, support tickets, app store reviews, in-app feedback, and sales calls. Most of this feedback gets lost in silos. Even when teams try to centralize it, manually categorizing and prioritizing hundreds of items per week is unsustainable. ProductSights automates this entire workflow with AI, so teams spend time acting on insights instead of sorting them.

### Who Uses ProductSights

- **Product managers** who need to prioritize features based on real user signals
- **Customer success teams** who want to surface recurring issues
- **Engineering leads** who want to understand top user problems directly from the IDE
- **Founders and product leaders** who need a data-driven view of what to build next

## Integrations

### Inbound Feedback Channels (12 total)
#### Configurable sources in Settings
- **Manual** — hand-entered feedback via the dashboard with templates (General, Sales, Support, Product)
- **Slack** — monitors specified channels for feedback messages (bot and webhook modes)
- **Intercom** — imports conversations and messages
- **Zendesk** — imports support tickets
- **Email** — dedicated inbox for forwarding feedback via Resend
- **App Store** — Apple App Store review ingestion
- **Google Play** — Google Play Store review ingestion
- **MCP (Analytics Tools)** — connect Amplitude, Mixpanel, or other MCP-compatible tools

#### Direct capture channels
- **Widget** — lightweight embeddable in-app feedback capture component (@productsights/widget); supports floating button, headless (trigger from your own UI), and embed (inline form) modes, with optional name/email/category fields and automatic page URL/title context
- **Browser Extension** — Chrome extension for capturing feedback from any webpage (@productsights/browser-extension)
- **Feedback Portal** — branded public pages at productsights.io/feedback/<org-slug> for end-user submissions
- **API Capture** — direct submissions to `/api/public/capture` from custom clients using an API key

### Outbound Destinations (Insight Sync)
- **Jira Product Discovery** — syncs categorized insights as ideas/opportunities (OAuth 2.0 and API token auth)
- **Productboard** — syncs insights with notes and metadata
- **Webhooks** — sends insight data to any HTTP endpoint for custom integrations

### Analytics Enrichment
- **Analytics Integration** — connect Amplitude, Mixpanel, PostHog, or any MCP-compatible analytics tool to enrich insights with usage statistics
- ProductSights maps product areas from feedback to analytics events (AI + manual overrides)
- Surfaces DAU, daily volume, and trends in the insight detail view; high-usage features get a priority boost
- Configure in Settings → Analytics; event catalog syncs daily; manual refresh available per insight or bulk

### Developer Tools
- **MCP Server** — connects to AI coding tools like Claude Code, Cursor, VS Code, and Windsurf, allowing developers to query product insights, top user problems, and sentiment data directly from the IDE
- Available via: `npx @productsights/mcp-server`
- Exposes tools for search, top problems, analytics stats, recent/related insights, weekly decision packets, and cluster impact lookup; cluster detail in the app adds evidence, discovery workspace, files, and analytics beta when configured
- **API key onboarding guard** — API key owners must select a role/department before creating keys or using widget, browser extension, or API capture flows
- **Campaign attribution support** — preserves first-touch `gclid` and UTM parameters through landing + auth and can emit Google Ads signup conversions when IDs are configured

## AI Triage Pipeline

Each piece of feedback is processed through an AI pipeline that:
1. **Categorizes** — classifies as bug report, feature request, praise, complaint, question, or other
2. **Sentiment Analysis** — determines positive, negative, neutral, or mixed sentiment
3. **Priority Scoring** — assigns urgency level (critical, high, medium, low)
4. **Summary Generation** — creates a concise one-line summary
5. **Entity Extraction** — identifies product areas, features, and key themes mentioned
6. **Deduplication** — detects and merges semantically similar feedback items

Results go into a **Review Queue** where product managers can approve, reject, or adjust the AI's classification before syncing to roadmap tools.
Review decisions support override-reason taxonomy and feed a continuous trust loop (confidence-band monitoring + prompt/rule tuning recommendations).

### AI-Powered Analytics

The analytics dashboard includes a natural language querying feature. Product teams can ask questions like:
- "What are the top complaints this month?"
- "Show me all feature requests related to onboarding"
- "What's the sentiment trend for mobile users?"

The AI draws from triaged insights to provide data-backed answers.
The dashboard also includes value-delivery metrics (activation and throughput), weekly decision packets for PM/engineering planning, and proactive alert cards for high-signal events.

### Insight clusters (product opportunities)

Each cluster has a detail page oriented around **product discovery**: ranked **customer evidence** (quotes across companies), a **signal** strip (volume, sentiment, priority), optional **analytics beta** context when an analytics MCP is configured, **discovery** fields (problem, hypotheses, open questions, research notes, next step), and **research file** attachments (images, PDF, video, notes). The list view supports filtering by **theme** via `?theme=<themeId>` on the clusters URL (e.g. from analytics top themes). Edits can lock AI-generated title/summary when appropriate so triage jobs do not overwrite PM copy.

## Public Feedback Portals

Organizations can create branded feedback portals at `productsights.io/feedback/<org-slug>`. These portals:
- Accept user feedback with optional name and email fields
- Are customizable with brand colors, headings, and descriptions
- Support short links (`productsights.io/f/<code>`) for sharing
- Feed directly into the AI triage pipeline

## Pricing

| Plan | Price | Sources | Insights/Month | Seats | Key Features |
|------|-------|---------|----------------|-------|--------------|
| Free | $0/mo | 1 | 100 | 1 | Manual sync only |
| Pro | $49/mo | 5 | 2,000 | 3 (+$15/extra) | 1 sync integration, priority email support |
| Team | $149/mo | 15 | 10,000 | 10 (+$12/extra) | Unlimited sync integrations, advanced rules engine, custom field mapping |

## Competitive Alternatives

ProductSights is an alternative to tools like Productboard, Canny, UserVoice, and Dovetail. Key differentiators:
- AI-first triage (not manual tagging)
- Native MCP server for IDE integration
- Ingests from 12 source types out of the box
- Affordable pricing starting at $0/month

## Technology

- Built with Next.js, React, TypeScript, and Tailwind CSS
- PostgreSQL database with Prisma ORM and pgvector for semantic search
- AI pipeline powered by large language models (OpenAI)
- Background job processing with BullMQ and Redis
- Deployed on Railway

## Frequently Asked Questions

### Which tools do you integrate with?

We support 12 feedback input channels: 8 configurable source integrations in Settings (Manual, Slack, Intercom, Email, Zendesk, App Store, Google Play, and MCP analytics tools) plus 4 direct capture channels (the feedback widget, browser extension, feedback portal, and API capture). We sync insights to Jira Product Discovery, Productboard, and custom webhooks.

### How does the AI triage work?

Our AI pipeline analyzes each piece of feedback to determine its category (bug, feature request, etc.), sentiment, and priority. It also generates a concise summary and extracts key entities like product areas.

### Is my customer data secure?

Yes. We take security seriously. Data is encrypted in transit and at rest. We are working towards SOC 2 compliance. We never use your data to train public AI models.

### Can I manually review the AI suggestions?

Absolutely. ProductSights includes a 'Review Queue' where you can approve, reject, or adjust the AI's classification and priority before syncing anything to your roadmap tools.

### How does the IDE integration work?

ProductSights ships an MCP server (npx @productsights/mcp-server) that exposes five tools — search insights, get top problems, view analytics stats, fetch recent insights, and find related feedback. Connect it to Claude Code, Cursor, VS Code, or Windsurf and your team can query product intelligence without leaving the editor.

### Can I use any MCP server as a data source?

Yes. ProductSights can connect to any MCP-compatible server as a feedback source. You configure the server command, tool, and output mapping — and every item flows through the full AI triage pipeline including classification, sentiment analysis, and deduplication.

### Can I enrich insights with analytics data?

Yes. Connect Amplitude, Mixpanel, PostHog, or any MCP-compatible analytics tool in Settings → Analytics. ProductSights maps product areas from feedback to your analytics events and surfaces usage stats (unique users, daily volume, trends) in the insight detail view. Insights affecting high-usage features get a priority boost.

### What is the feedback widget?

The feedback widget is a lightweight, embeddable component you can drop into your web app to capture user feedback in context. It supports three modes: a floating button (default), headless mode (no button — trigger it from your own UI), and embed mode (render the form inline in any container). There is also a Chrome browser extension for your team to capture feedback from any webpage, and branded public feedback portals for collecting submissions from end users. All three route directly into your ProductSights pipeline.

### Can I ask questions about my feedback data?

Yes. The analytics dashboard includes an AI querying feature — type a natural language question like "What are the top complaints this month?" and get an instant, data-backed answer drawn from your triaged insights.

### What are insight clusters?

Clusters group similar feedback automatically using embeddings so you can see recurring themes. Each cluster has a product opportunity view: customer evidence quotes, signal strength, a discovery workspace (problem, hypotheses, notes), optional analytics context when connected, and research file attachments. You can filter the cluster list by theme from analytics.

### What happens if I exceed my plan limits?

We'll notify you if you're approaching your limit. If you exceed it, we'll continue collecting feedback but pause the AI processing until the next billing cycle or an upgrade.

## Links

- Website: https://productsights.io
- Documentation: https://productsights.io/en/docs
- Browser Extension: https://productsights.io/en/browser-extension
- Blog — Why product insights belong in your IDE: https://productsights.hashnode.dev/why-product-insights-belong-in-your-ide
- Concise LLM reference: https://productsights.io/llms.txt
- ProductSights vs Productboard: https://productsights.io/vs/productboard
- ProductSights vs Canny: https://productsights.io/vs/canny