Organizing Workspace Topics

Topics are how Agency Hero organizes everything you discuss across time. When meetings are processed, the AI extracts the themes that came up and maps them to the workspace’s canonical topic registry. Over time, a topic like “API Migration” or “Q2 Deliverables” accumulates every decision, risk, open question, and action item that has touched it — across every meeting in which it appeared.

This article covers how topics work, why you should set them up before meetings begin, how post-meeting matching keeps them current, and how the Topic Brief turns that accumulated knowledge into a live, readable summary.

Start with topics before your first meeting#Copy link

The single highest-leverage thing you can do when setting up a new workspace is to create topics before any meetings are processed.

Topics are the organizing structure that the AI uses to route intelligence. When the system processes a meeting and extracts a decision or risk, it links that item to the closest matching canonical topic. If no topics exist yet, items either fall into a catch-all topic or trigger the creation of new topics that may not align with your actual workstreams.

By pre-creating topics that map to your project’s scope of work — key deliverables, initiatives, workstreams, client concerns — you give the AI an intentional taxonomy to match against from day one.

Best practice: When you kick off a new project, open the Knowledge → Topics tab and create topics for each major area of work before your first meeting. Think in terms of: What are the recurring themes this engagement will produce? What subjects should we be able to filter and review independently?

Good topic names are short, meaningful, and stable. Examples:

• Budget Alignment
• Technical Integration
• Q3 Rollout Plan
• Stakeholder Sign-off
• Risk: Data Migration

Avoid creating topics that are too granular (“Monday standup items”) or too broad (“Everything”). Topics work best when they map to a real, recurring thread you expect to monitor and revisit over weeks or months.

How to create a topic manually#Copy link

Navigate to your workspace’s Knowledge → Topics tab. Use the New Topic button to open the creation dialog. Enter the topic name and confirm. The topic is immediately registered in the workspace’s canonical topic registry and is available for AI matching on the very next meeting that is processed.

Manual topics start with a discussion_count of 0 — that’s correct behavior. As meetings contribute to the topic, the count climbs and the topic gains presence in the workspace’s context.

How post-meeting topic matching works#Copy link

When a meeting is processed, the system runs a multi-step topic matching pipeline:

1. Topic extraction — The AI reads the transcript and extracts the discrete themes that were discussed in that meeting. These become meeting topics, each with a name, key points, and any decisions, action items, or concerns specific to that discussion.
2. Candidate retrieval — For each extracted meeting topic, the system builds a candidate set from the workspace’s existing canonical topics. This uses two signals in combination:
   • Exact match — A normalized version of the meeting topic name (lowercased, punctuation stripped) is compared against the existing canonical topic registry.
   • Semantic match — Embedding similarity is used to find canonical topics that are conceptually close, even if the wording differs.
3. LLM match proposal — An LLM step evaluates the candidates and makes a recommendation: link to an existing canonical topic, create a new one, or flag for manual review. The recommendation includes a confidence score and a short rationale.
4. Human confirmation — The system does not silently apply matches. High-confidence matches are pre-selected in the review UI, but you always confirm. Lower-confidence matches are surfaced for manual selection or creation.

This pattern — retrieve, reason, propose, confirm — is consistent with how Agency Hero handles all AI-extracted intelligence. Nothing enters the canonical workspace record without a human in the loop.

When a new topic is created#Copy link

If no existing canonical topic is a good match for a meeting theme, the system can propose creating a new one. You can accept the proposal (which registers the new topic) or redirect the meeting topic to an existing canonical topic by searching and selecting manually.

You can also create a canonical topic at any time from the Knowledge → Topics tab and then link existing meeting topics to it retroactively — useful when you realize a topic needs to be split, renamed, or reorganized.

Merging topics#Copy link

Over time, you may end up with two topics that represent the same thread under slightly different names — for example, “API Integration” and “Integration Architecture”. Agency Hero supports canonical topic merges: one topic is designated the winner, the other is archived, and all historical meeting contributions, intelligence items, and provenance rows are repointed to the winner. The full timeline remains intact.

Managing your topic list#Copy link

Create#Copy link

From Knowledge → Topics, click New Topic, enter the name, and confirm. The topic is added to the registry immediately and will be available for AI matching on the next processed meeting.

Rename#Copy link

You can rename any topic directly from the topic detail page. Renaming updates the canonical name and re-normalizes it for matching. If the new name collides with an existing topic, Agency Hero will prompt you to merge instead of silently overwriting.

Archive#Copy link

Topics that are no longer active can be archived. Archived topics are hidden from the default topic list but retain their full history and can be restored. They will not be proposed as matches for new meeting topics.

Merge#Copy link

When two topics should have been one all along, merge them. Select the winner topic, merge the other into it, and all contributions, intelligence items, and briefs are consolidated.

Tip: Keep your topic list clean and focused. A workspace with 5–10 well-named topics produces much better AI-matched intelligence than a workspace with 40 overlapping micro-topics. When in doubt, merge.

The Topic Brief#Copy link

Every canonical workspace topic has a Topic Brief — an AI-written summary that synthesizes everything known about that topic from across all meetings and linked sources.

What the Topic Brief contains#Copy link

The brief is generated from three context providers:

• Meeting contributions — All meeting topics linked to this canonical topic, up to the most recent 10. Each contribution includes the meeting-scoped topic text, key points discussed, and any details extracted in that session.
• Open intelligence items — Open decisions, open risks, and unanswered questions that have been confirmed and linked to this topic (up to 10 per type).
• Open tasks — Incomplete tasks that originated from a meeting topic associated with this canonical topic (up to 20).

The result is a coherent narrative summary of where this topic stands: what has been decided, what’s at risk, what’s still unresolved, and what tasks are in flight.

Example — what a brief for “Data Migration” might show:

• Summary of the approach agreed on across three meetings
• Two open risks still flagged (data schema mismatch, third-party API rate limits)
• One unanswered question about the cutover window
• Three tasks in progress with owners

When the brief regenerates#Copy link

Topic briefs regenerate automatically when the underlying content changes. The triggers that initiate a regeneration include:

• A meeting topic is newly linked to the canonical topic
• A new intelligence item is linked to the topic
• The topic itself is created or modified
• A manual regeneration is requested by the user

A cooldown (default: 60 seconds) prevents excessive regeneration. If multiple changes arrive in rapid succession, they are coalesced — a single fresh brief is generated once the cooldown passes.

The brief is stored with a generated_at timestamp so you always know how recent it is.

What the Topic Brief is not#Copy link

The brief is derived, not authored. You cannot manually edit it. It reflects the current state of linked content — when that content changes, the brief changes with it.

If you want a stable, human-authored narrative about a topic — a strategy note, a client update, a retrospective — create a Memo instead. You can seed a memo directly from the current Topic Brief snapshot and then edit it. The memo is yours; it won’t auto-update.

How to use the Topic Brief#Copy link

• Before a meeting — Pull up the relevant topic brief to get a quick, current picture of where things stand. Share it with team members as a pre-read.
• For status updates — Use the brief as a starting point for client or stakeholder updates. The AI has already synthesized the key points; you can write a memo from it.
• For onboarding — When someone new joins a project mid-engagement, the topic briefs give them a fast, structured catch-up on each workstream without reading through transcripts.
• For reviews — Scan your topic list regularly. Topics with open risks and unanswered questions that haven’t had a meeting contribution in two weeks are worth scheduling time around.

Keeping your topics healthy#Copy link

A workspace’s topic structure is a living taxonomy. Here are practical habits that keep it working well:

Seed topics at kickoff. As described above, create the expected topics before your first meeting. Even a rough set of 4–6 topic names will dramatically improve first-meeting matching quality.

Review new topic proposals after each meeting. During your post-meeting review, check whether the AI has proposed any new canonical topics. Accept the ones that represent genuine recurring threads; redirect or dismiss the ones that are one-off items.

Merge opportunistically. When you notice two topics that mean the same thing, merge them. Each merge improves future matching because the canonical registry becomes less ambiguous.

Archive completed workstreams. When a deliverable is done or a workstream closes, archive the topic. This keeps your active topic list focused and prevents stale topics from attracting new matches.

Watch discussion_count. Topics with a high discussion count are the core threads of your engagement. Topics with count 0 or 1 that have sat untouched for several weeks either need to be promoted into your next agenda or archived.

Topics in the workspace context#Copy link

Topics are not just a list — they are the organizational backbone of the workspace’s intelligence layer. When Agency Hero assembles a context pack to power AI chat, workflow agents, or brief generation, it draws on the canonical topic registry as part of its understanding of what this workspace is about.

This means that a well-organized topic structure doesn’t just help your team navigate knowledge — it directly improves the quality of every AI-assisted action in the workspace. The AI knows what “Data Migration” or “Q3 Deliverables” means in the context of this specific engagement because those topics exist, carry linked intelligence, and have a brief that describes the current state.

Setting topics up early, maintaining them consistently, and confirming post-meeting proposals as they arrive is the foundational practice for getting the most out of Agency Hero’s intelligence layer.