> ## Documentation Index > Fetch the complete documentation index at: https://modem.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # Topics > How Modem discovers and organizes conversation themes **Topics** are AI-discovered conversation themes that group related messages across your data sources. They're the primary way Modem helps you understand what users are talking about. ## How Topics Are Created As messages arrive from your connected sources, Modem analyzes them and groups related conversations into topics. When a conversation fits an existing topic, it's added to that topic. When it describes something new about your product, Modem creates a new topic. Modem skips conversations that aren't about your product, such as logistics chatter, bot notifications, or internal ops, so they never become topics. A conversation can raise more than one concern, so a single conversation can contribute to several topics. Each topic keeps track of the conversations it was built from (see [Sources](#sources) below). Topics are generated automatically. You don't need to tag, label, or categorize messages yourself. ## What a Topic Contains Each topic has an AI-generated title and summary, along with a priority level and the people involved. ### Feedback Classification Each topic is classified by feedback type: | Category | What it captures | | - | - | | **Bug Report** | Error descriptions, broken functionality, unexpected behavior | | **Feature Request** | Suggestions, wishlist items, enhancement ideas | | **Complaint** | Frustration, dissatisfaction, negative experiences | | **Praise** | Positive feedback, compliments, appreciation | Not every topic fits one of these categories. Topics that don't are classified as **Discussion**: general conversation, coordination, and questions. You can filter Discussion topics in or out of the list. ### Summaries Modem generates summaries to help you scan topics quickly: * **Topic summary.** A short overview of what the topic is about and its key themes. * **Evidence timeline.** The messages behind the topic, in order, alongside the topic's change history, so you can trace any point in the summary back to its source. ## Sources Each topic lists the conversations, issues, and pull requests it was derived from under **Sources**. They can come from different channels and different times. For example, a topic about "API rate limiting" might draw on: * A Slack #support thread about rate limit errors (January) * A Discord #suggestions post asking for rate limit headers (March) * A GitHub issue about incorrect 429 responses (May) Sources help you see how a topic evolved over time and jump to the original conversation. ## Working with Topics ### Browsing Navigate to **Topics** in your dashboard to see discovered topics, sorted by most recent activity. Depending on your organization's settings, the list may open on a default view that hides some topics, such as Discussion and very-low-priority topics. Switch the view or clear those filters to see everything. Filter by: * **Search.** Find topics by keyword in the title. * **Type.** Bug report, feature request, complaint, praise, or discussion. * **Priority.** Very low through very high. * **Lifecycle.** Open, in progress, completed, dismissed, or unknown. * **People.** Topics a specific person has participated in. * **Category.** Filter by topic keywords or tags. * **Product.** Topics about a specific product, when your organization has products set up. * **Company.** Topics where a specific company's users have participated. * **Channel.** Topics from a specific source. * **Ticket status.** Topics with or without a linked ticket. * **Team response.** Whether your team has responded. Most filters can be switched between **is** and **is not** to include or exclude values. Some filters aren't available on every organization's topic list. ### Searching The search box matches keywords in topic titles. Wrap a word in double quotes to match that exact word, so `"persona"` doesn't match "personal." To find topics by meaning rather than exact words, ask the [Modem Agent](/docs/features/agent), for example "Which topics are about authentication problems?" ### Topic Details Click on a topic to see: * The summary and the evidence timeline of messages behind it * People involved and their contributions * Sources: the conversations, issues, and pull requests the topic was derived from * Products and keywords assigned to the topic * Priority and lifecycle state, which you can change * Change history ## Automations Topic activity can trigger [automations](/docs/features/automations). You can create trigger automations that run when: * **A new topic is created.** * **A topic reaches high priority.** Depending on how your organization scores priority, this fires at High or at Very high. * **A conversation is added to a topic, or gets a follow-up on one.** These events currently cover X sources, and the automation must be scoped to specific channels or products. Use these when you want Modem to react immediately, for example by posting an alert to Slack, creating follow-up work, or sending a summary to the right team. ## Related Unified profiles for users and organizations. Ask questions and take action using natural language. This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.