Searching Across Meeting History

Your meetings accumulate a wealth of context over time: decisions made, risks flagged, questions raised, and conversations that shaped your work. Searching across that history lets you surface any of it in seconds — without needing to remember exactly which meeting it came from or precisely how something was phrased.

Agency Hero’s meeting history search is semantic, which means it understands the meaning of what you’re looking for, not just matching keywords. Ask “what did we decide about pricing?” and you’ll find relevant decisions even if the word “pricing” never appears verbatim in the meeting summary.

How Semantic Search Works#Copy link

When a meeting summary is finalized, Agency Hero converts its content into a vector embedding — a mathematical representation of meaning. The same happens for each individual discussion topic, and for every confirmed decision, risk, and question extracted from the meeting.

When you run a search, your query is converted to an embedding in the same space, and the system finds content whose meaning is closest to yours. The practical effect:

• Plain-language queries work. “Timeline concerns from the client” will surface risks and topics about schedule pressure even if those exact words were never used.
• Synonyms and paraphrases are handled. “Budget approval” finds content about “cost sign-off” or “financial authorization.”
• Specificity helps. “What did we decide about quoting?” is more targeted than “quoting” and will more reliably return decision-level results.

If semantic search returns no results, the system automatically falls back to keyword matching so you still get relevant hits on verbatim text.

Why results vary by meeting: Meetings that cover many topics at once produce a more diluted summary embedding. Agency Hero addresses this by also embedding each individual topic block separately, so even a wide-ranging meeting will surface specific subtopics accurately.

What Gets Searched#Copy link

Search operates across four layers of meeting intelligence, each suited to a different type of question:

Note: Action items are intentionally excluded from semantic search. Because tasks change frequently — status updates, reassignments, due date edits — keeping them in the search index would produce noisy, stale results. Use the task list or filter views to find action items.

Searching via the AI Assistant#Copy link

The most powerful way to search your meeting history is through the Agency Hero AI assistant. You can ask questions naturally and the assistant will search the appropriate layer automatically.

To search via the assistant:

1. Open the AI assistant from any workspace (the chat icon in the sidebar).
2. Type your question in plain language.

Example queries:

• “What was discussed about the HubSpot quoting workflow?” — searches discussion topics
• “What decisions have we made about resourcing?” — searches confirmed decisions
• “What risks have we flagged related to data security?” — searches confirmed risks
• “Find meetings about the NetSuite integration” — searches full meeting summaries
• “What questions came up about timeline in the last 30 days?” — combines topic search with a date filter

The assistant selects the right search layer based on your question, then presents results ranked by relevance. If one layer returns nothing, it automatically tries the others.

Scoping the search#Copy link

By default, the assistant searches the current workspace. You can broaden or narrow scope explicitly:

• “Search across all workspaces” — expands to your entire organization
• “In the Acme Corp workspace, what risks have we flagged?” — scopes to a named workspace
• “Meetings about pricing in the last two weeks” — applies a date window

Searching from the Meetings Page#Copy link

The Meetings page provides a direct search bar and filter controls for browsing your meeting list by title and metadata.

Using the search bar#Copy link

1. Navigate to Meetings in the left sidebar.
2. Use the search bar (top-right in card or table view) to filter by meeting title or description.
3. Results update instantly as you type.

The meetings page search filters by title and description only — it is not a semantic search. Use the AI assistant for meaning-based queries across summaries, topics, and intelligence items.

Switching between views#Copy link

The Meetings page offers multiple views, each with the same search and filter controls:

• Table view — compact list, best for scanning many meetings quickly
• Card view — shows attendees, workspace tags, and platform details at a glance
• Calendar views (Day / Week / Month) — navigate by date to find meetings in a specific period

Use the Upcoming / Past toggle to switch between future and historical meetings.

Filtering by meeting type#Copy link

Click Filter to narrow results by meeting type:

• Sales — prospect and deal-related meetings
• Internal — team and planning meetings
• Client — active client delivery meetings

Multiple types can be selected at once. The active filter count is shown on the button so it’s always clear when filters are applied.

Available Search Filters#Copy link

When searching via the AI assistant, the following filters can be applied in natural language:

Tips for Better Search Results#Copy link

Be specific about what you’re looking for. “What decisions did we make about the data migration timeline?” will outperform a single-word query like “migration.”

Name the type of intelligence you need. If you know you’re looking for a decision vs. a risk vs. a question, say so. The assistant uses this to narrow to the right search layer.

Use natural sentence structure. Queries like “what did we agree on regarding pricing discounts” work just as well as keyword-style queries — often better.

Add a time context when relevant. If you remember roughly when something happened, narrow the window: “in the last two months” or “before the Q3 kickoff.”

Try different phrasings if results are thin. If a meeting is recent and its summary hasn’t been finalized yet, it won’t surface. Likewise, intelligence items only appear after they’ve been confirmed during post-meeting review.

Confirm intelligence items to make them findable. Decisions, risks, and questions only appear in search after a team member has reviewed and confirmed them. Unconfirmed (proposed) items are excluded to keep search results reliable.

Limitations and Scope#Copy link

Coverage depends on finalization#Copy link

A meeting summary must be finalized before it becomes searchable. Finalization happens in one of two ways:

• A team member clicks Finalize Summary on the meeting detail page.
• The system auto-finalizes the summary 48 hours after the meeting ends if no one has done so manually.

Until then, the meeting won’t appear in semantic search results.

Intelligence items require confirmation#Copy link

Decisions, risks, and questions are only embedded and indexed after they are confirmed during post-meeting review. Items left in the “proposed” state (AI-generated but unreviewed) are not searchable. This is by design — it keeps search results reliable and prevents unverified extractions from appearing in your organization’s knowledge base.

Tasks are not semantically searched#Copy link

Action items and tasks are not included in semantic search because they’re mutable work items that change constantly. Search for tasks using the task board, assignee filters, or by asking the assistant to list tasks with structured criteria (e.g., “tasks assigned to me this week”).

Privacy and visibility#Copy link

Your signed-in identity is always used to enforce access rules. Meetings marked as Private or Restricted are only visible to authorized attendees and admins — this applies in search results as well. You will never see a search result for a meeting you don’t have access to.

Cross-workspace search requires explicit opt-in#Copy link

By default, all searches are scoped to your current workspace. To search across all workspaces in your organization, explicitly ask for it (e.g., “search all workspaces”). This prevents results from unrelated client projects cluttering your answers.

Related Articles#Copy link

• Post-Meeting Process — How to review and confirm intelligence items so they become searchable.
• Meeting Summaries — How meeting summaries are generated and finalized.
• Understanding Intelligence Items — What decisions, risks, and questions are and how they’re extracted.
• Meeting Topics — How discussion topics are structured and what makes them searchable.
• Understanding Meeting Intelligence Capture — The full end-to-end picture of how meetings become structured knowledge.