Follow an idea across conversations.
The problem
A summary helps with one conversation. An archive of separate summaries still makes it hard to see how a project, a decision and a later change fit together.
An explanatory exhibit
Three conversations, one project
Illustrated extraction with invented records. These are explanatory examples, not product output.
- 01
Conversation
“We chose SQLite for the Lantern prototype.”
- 02
Relationship
Lantern / uses / SQLite
- 03
Entity
Lantern keeps a link to that source
Inspect the consequence
Follow the project
Lantern / SQLite
A project node joins mentions from separate conversations. The relationship retains its source.
Check the source
Read the original statement
The invented sentence supports the prototype decision. It does not establish that every later release used SQLite.
What this makes possible
I extended conversation analysis with entities, relationships and dated events, so an idea could lead back to the conversations that explain it.
Watch the actual graph workflowRecorded in the product
Focus on a project and its connections
What happens in this recording
Actual PanelForge interface · public-safe project and software relationships · edited recording
The real graph is filtered to PanelForge and its build relationships. Selecting a project brings connected technologies into focus; zoom and fit change the view. These are extracted relationships, including historical names and aliases, rather than a verified inventory of the current architecture.
Source information sheet
Evidence, implementation & history Implementation inspected
What I checked
Analysis and graph source inspected. A fresh, public-safe recording verifies graph search, predicate filtering, focus, zoom and fit. Other analysis behavior remains supported by source inspection.
Research cut: 14 October 2026. Source areas below are from panel-forge. Private conversation references are retained in the research ledger and are not published here.
Implementation decisions
The analysis pipeline stores structured relationships as knowledge triples. Entity and event information gives those relationships a shared vocabulary and a place in time.
I added graph views and entity details alongside the archive. Following a connection can lead to the supporting conversation, where the original wording and context are still available.
A path through the feature
- Begin with an entity
Choose a project or idea in the graph, then inspect its relationships.
- Follow the evidence
Open a linked conversation to check what was actually said. An extracted relationship is a useful lead that still needs its source.
Source areas
app/Services/ConversationAnalysisService.phpapp/Jobs/AnalyzeConversationJob.phpapp/Models/KnowledgeTriple.phpresources/js/pages/KnowledgeGraph/
The small explanatory exhibit uses invented records to describe the mechanism. Separate recordings and working examples carry their own evidence notes.
Recorded history
Dates describe the researched record, rather than today’s deployment.
Building conversation analysis and a personal knowledge graph
Implementing knowledge graph extraction