Most meeting tools give you a flat archive — a list of transcripts you scroll through to find something. A knowledge graph gives you a connected web of everything ever said, decided, or committed to across all your meetings. The practical difference is dramatic: instead of searching for a transcript, you navigate a visual map of your entire meeting history.
What is a knowledge graph?
A knowledge graph represents information as nodes (things) and edges (relationships between things). Unlike a flat database where you search for records, a knowledge graph lets you traverse relationships. In meetings, nodes are: people, topics, decisions, and action items. Edges are the relationships between them — "Sarah raised the budget concern in the Q3 review, which led to the decision to delay the launch." The result is a navigable map of institutional memory, not just a text archive.
The problem with flat meeting archives
A flat archive has one search mode: keyword. Type "budget" and get a list of transcripts mentioning the word. You then read each one to find the relevant context. With 50 meetings, this is manageable. With 500, it breaks down completely. The deeper problem is that context is relational. A decision made in January connects to a conversation in March, which connects to an action item in May. Flat search finds the word; it cannot surface the thread. Most teams lose this context entirely — it evaporates into transcripts that nobody re-reads.
How a meeting knowledge graph works
AI identifies named entities in the transcript: people, organisations, products, dates, amounts. These become nodes.
Related sentences and paragraphs are grouped by theme. "Q3 roadmap," "pricing," and "hiring plan" become topic nodes.
AI identifies how entities relate — "Sarah approved the Q3 roadmap" creates edges between Sarah, the approval event, and Q3 roadmap.
When the same entity appears in multiple meetings, the graph links those meetings. Now you navigate from a person to every meeting they appeared in, from a topic to every time it was discussed, or from a decision to the action items it generated.
In Wisprnote AI, this graph is rendered as an interactive force-directed visualisation — a web of interconnected nodes you can drag, zoom, and click.
5 benefits over flat transcript search
Search for "pricing" and get only the meetings where pricing decisions were made — not every transcript that mentions the word.
Click a person to see every meeting they appeared in, every topic they raised, every action item assigned to them.
See when "technical debt" first appeared, how often it's discussed, and what decisions have been made each time.
A new hire explores the knowledge graph to understand project history — who decided what, when, why — without reading every transcript from the past year.
Wisprnote AI's AI chat is connected to the knowledge graph. Ask "What commitments did we make to Acme Corp in the last 90 days?" and get an answer with citations.
Wisprnote's knowledge graph in practice
The graph is built automatically from every meeting recorded — no setup required. Practical examples: click a person's name to see every meeting they appeared in; click a topic to see the chronological thread of conversations; click an action item to see the meeting where it was created and whether it was completed; ask AI chat "What did we decide about enterprise pricing?" and get a sourced answer from the graph.
Frequently asked questions
A meeting knowledge graph represents the people, topics, decisions, and action items from your meetings as linked nodes. Instead of searching a flat transcript list, you navigate a visual web of relationships — clicking a person to see their meeting history, a topic to see its thread, or querying the entire corpus in plain English.
Full-text search returns transcripts containing a keyword. A knowledge graph returns entities related to a concept with their relationships. Search "budget" in full-text and get every transcript mentioning the word. The knowledge graph returns the budget discussions, who was involved, what decisions were made, and what action items they generated — with connections visible.
It becomes useful from 5–10 meetings and valuable from 20+. With fewer meetings, the graph is sparse but still useful. The more meetings, the richer the connections and the more powerful the cross-meeting queries become.
Yes. On Wisprnote AI's Team plan, the knowledge graph spans all meetings shared in a team workspace — so a sales discovery call and a product roadmap session can be connected if they reference the same client, feature, or commitment.
Wisprnote uses named entity recognition, topic clustering, and LLM analysis to extract nodes and relationships from every transcript. The process runs automatically after each meeting ends and completes within 60 seconds.