Knowledge Graph

Caution

This feature is available for customers with the Knowledge module.

The Knowledge Graph adds a structured layer to the Knowledge Store: StoryVault automatically extracts decisions, tasks, people, and relationships from your sessions. Instead of just searching for text passages, you can ask specifically: “Which tasks were assigned to Ms. Müller?” or “Which decisions depend on the new care concept?”

What Gets Extracted?

The AI recognizes the following elements in your transcripts:

Element	Description	Example
Decisions	Resolutions made	“We will switch to the new documentation system from Q2”
Tasks	Action items and responsibilities	“Mr. Schmidt will review the offers by Friday”
Ideas	Suggestions and proposals	“We could introduce a mentoring program”
Insights	Important findings	“Occupancy in day care has been declining for three months”
Milestones	Achieved goals	“The new QM handbook is complete”
Changes	Documented changes	“The duty roster was switched to 12-hour shifts”
People	Mentioned and responsible persons	Speakers, responsible parties, participants
Topics	Specialist areas and focal points	Care quality, staffing, hygiene

These elements are linked to each other: a task is assigned to a person, a decision depends on another, a topic connects multiple sessions.

Extracting Knowledge from a Session

1. Open a session

   Navigate to a session with a completed transcript.

2. Click “Extract Knowledge”

   

   The extraction runs in the background and takes a few seconds to minutes depending on transcript length.

3. View the result

   After processing, you’ll see the knowledge graph of the session: decisions, tasks, people, and their relationships at a glance.

Tip

Automatic extraction: Whether knowledge is extracted automatically after transcription depends on the project configuration. If the graph is missing, first check processing status and the available manual extraction action. Ask your administrator about configuration; the current project form has no dedicated switch for it.

Project-Wide Knowledge Graph

The real value emerges when you view the knowledge graph at project level. StoryVault automatically links people and topics across session boundaries:

• If “Ms. Müller” is mentioned in three different sessions, the system recognizes it’s the same person
• Topics like “care quality” connect all sessions where this topic was discussed
• Tasks from one session can be recognized as completed when referenced in a later session

How to access the project knowledge graph:

Navigate to the project and open the Knowledge Graph section. You’ll see all extracted elements from all sessions in the project — linked together.

Batch Extraction

If you have a project with many existing sessions, you can start the extraction for all sessions at once:

Project page → “Extract All Sessions”

The processing runs in the background. Depending on the number of sessions, this may take several minutes.

Knowledge Graph in AI Chat

The Knowledge Graph is automatically used in AI chat. When you ask a question in session or project chat, the AI incorporates the structured information. This leads to more precise answers because the AI doesn’t just find text passages — it understands relationships.

Example questions that benefit from the Knowledge Graph:

• “What open tasks are there currently in this project?”
• “Who is responsible for the topic of occupational safety?”
• “What decisions were made in the last three sessions?”
• “Are there tasks that depend on the budget decision?”

Difference from the Knowledge Store

	Knowledge Store	Knowledge Graph
Search	Text-based — finds relevant text passages	Structured — finds decisions, tasks, people
Sources	Any (sessions, texts, files, lessons)	Automatically from session transcripts
Strength	Broad knowledge base, many source types	Relationships and connections
Use	Looking up knowledge	Tracking decisions and tasks

Both systems complement each other: the Knowledge Store provides the broad text base, the Knowledge Graph provides the structured relationships.

Tip

Practical Example: A youth welfare provider uses the Knowledge Graph for their weekly case conferences. After six months, the team leader can ask: “What measures have been discussed for the Meier family in the last sessions?” — and receives a chronological overview of all decisions and tasks, linked across sessions.

Further Reading

• Knowledge Store →
• Add Sources →
• Use in Chat →

---

This page describes the Knowledge Graph (GraphRAG) — structured knowledge extraction from meeting transcripts.