Memory & Knowledge Graph
A personal Knowledge Graph built from your chats, files, meetings, and notes — searchable by every AI model, always growing, always learning.
The longer you use metir, the smarter it gets.
Context assembled from 4 sources for this conversation
All context injected into AI system prompt
How Memory Works
Your Knowledge Graph
metir builds a personal Knowledge Graph from everything you interact with — chats, uploaded files, meeting transcripts, and notes. Every piece of information is indexed, connected, and searchable by AI.
Head of marketing at Series B startup
Extracted from 12 conversations
Prefers bullet-point style responses
Extracted from 8 conversations
Working on pricing page redesign
Extracted from 3 conversations
Ingestion Pipeline
When you upload a file, record a meeting, or write a note, metir automatically processes it through a multi-stage pipeline — chunking, embedding, entity extraction, and relationship mapping — so AI can find exactly what it needs.
Upload or Record
Drop files, record a meeting, or create a note
Chunk & Embed
Content is split into semantic chunks and vector-embedded for search
Extract Entities & Relationships
People, companies, concepts, and how they relate are mapped
Generate Summaries
Document, cluster, and project-level summaries are created
Smart Retrieval
When you send a message, metir searches across all your knowledge sources in parallel — memories, document chunks, entities, relationships, and past conversations — then assembles only the most relevant context for AI.
Your message
“Help me prep for the board meeting”
Parallel search across all sources
Context assembled
Cross-Model Memory
Your Knowledge Graph works across every AI model. Upload a file while chatting with one model, switch to another — it already has access to the same documents, memories, and context. No re-uploading, no re-explaining.
“Summarise the key findings from the Q4 report I uploaded”
“Draft a board presentation based on Q4 results”
Source Attribution
Every AI response shows which sources were used — files, notes, meetings, and conversations. Click any source to jump directly to the original, so you can verify and explore further.
Privacy & Control
Your Knowledge Graph is private, encrypted, and protected by row-level security. Use incognito mode to chat without any context, or manage exactly what's stored.
Encrypted Storage
All data encrypted at rest and in transit
Row-Level Security
Only you can access your data — enforced at the database level
Incognito Mode
Chat without any context — nothing is stored or retrieved
Full Control
View, delete, and manage your files, notes, and memories
