RelationTrace Private investigation workspace

Reconstruct what happened from your archive.

Upload your archive export. RelationTrace rebuilds the timeline, detects relationship phases, and lets you investigate the evidence you couldn't see while living it.

01

Private by default

Your archive uploads to a private workspace. Nothing is shared or stored externally — the data is entirely yours.

02

Every conclusion traces to evidence

Findings link back to real messages, timestamps, and conversation context. You can verify every claim against the source.

03

Investigate iteratively

New questions surface as you go. Rerun enrichment, shift your focus, and refine your understanding over multiple passes.

Platform

From archive export to structured evidence.

RelationTrace turns raw conversation history into a navigable investigation workspace — people, timelines, phases, and evidence you can revisit.

People

Understand who mattered and when.

Browse contacts ranked by significance, explore their relationship trajectory, and drill from summary into the underlying messages.

Periods

See the phases you lived through.

Relationship phases — warming, tension, reconnection, silence — are detected from conversation patterns and surfaced as navigable timeline segments.

Investigations

Ask a question. Get evidence, not speculation.

Turn questions into trackable cases with pinned evidence, confidence scores, and follow-up paths grounded in your actual messages.

Enrichment

Layer context onto every conversation.

Sentiment, emotion, relational signals, and key events are detected across your archive and can be rerun as your questions evolve.

Graph & Context

Trace connections across people and time.

Interactive graph views scoped by person, period, or investigation — useful when you need to see how relationships overlap.

Privacy

Private by design. Your history stays in your workspace.

Designed for people who need structure and control when revisiting sensitive personal history on their own terms.

Workflow

Four steps from archive to investigation.

1

Upload your archive export

Import your archive ZIP from Settings. Facebook exports are currently supported, with more sources planned.

2

Rebuild the timeline

Messages, reactions, friends, and events are parsed into a structured, searchable history with people and conversations linked.

3

Enrich with context

Sentiment, emotion, relational signals, and key events are layered onto every conversation automatically.

4

Investigate the evidence

Browse people, explore relationship phases, pin evidence, and run structured investigations with follow-up questions.

Approach

Structured for clarity, not just search.

Instead of dumping everything into one view, RelationTrace separates people, relationship phases, evidence, and investigations so the timeline becomes easier to follow.

Traceability

Every conclusion links back to real messages.

Move from a finding to the actual conversation moments, event markers, and timeline shifts that support it — nothing is asserted without evidence.

Focus

Stay scoped. Pick up where you left off.

Investigations stay anchored to a person, conversation, or period so you don't rebuild context from scratch each session.

Iteration

Rerun analysis as your understanding changes.

The workspace supports iterative review — new enrichment passes, refined questions, and evolving perspectives over time.

Design principle

This isn't about more data. It's about seeing what was already there.

RelationTrace emphasizes evidence-backed reasoning: identify the right people, isolate the right phases, trace the evidence, and keep your investigation visible as you go.

Ready to start

Trace the evidence. Understand the timeline.

Request access to start your private investigation workspace, or sign in if you already have one.

Good morning

Loading...

Activity

Loading activity...

Connections Explore people →

Loading graph...

Archive
—
Messages
—
People
—
Conversations
Loading coverage...
▾
Embeddings
—
Sentiment
—
Emotions
—
Analysis
—

People

Constellation →
← People

← Periods

— Messages
— Participants
— Intensity
— Type

Summary

Active People

Events

Investigations · …

Ask a question. Get evidence, not speculation.

Open a new case or resume one in flight. Every answer is anchored to excerpts, counterevidence, and a confidence score — never just a vibe.

Scope
Confidence distribution last 40 answers
Recent —
Investigations › — · —

Using Case context

Settings

workspace — · Import data, configure AI pipelines, and manage your workspace.
Services healthy

Archive intake

Upload a Facebook or Messenger export. RelationTrace parses the archive, extracts people and conversations, then schedules enrichment in the background.

accepts .zip · up to 4 GB

Drop a Facebook or Messenger export ZIP

Files are parsed locally, hashed, and journaled. Duplicate archives are detected by SHA and merged into existing data.

facebook-export.zip messenger-archive.zip json · html max 4 GB

Enrichment

Run AI analysis on imported messages. Start these after an import completes.

Generate embeddings

Enables semantic search across messages.

Analyze sentiment & emotions

Per-message sentiment labels and emotion classification.

Language Settings
Configure interface and analysis language preferences.
Controls interface labels and text
Hints the LLM pipeline about message language
Workspace Privacy
Redacts PII (names, emails, phone numbers) from chat prompts before they reach OpenAI, and restores the original values in the response locally. Chat completions only — embeddings are unaffected.
Model Configuration
Choose which models are used for each pipeline stage. Changes apply to future analysis runs.
Embedding
—
Vector embeddings for semantic search
Fast Structured
—
Extraction, classification, explanation
Reasoning
—
Inference, state transitions, investigation, skill escalation
Presentation
—
Coaching, rewrite, explanation
API Keys
Manage credentials for external services.
Used for embeddings, sentiment, emotion, and analysis pipelines
Pipeline Parameters
Fine-tune analysis behavior.
Token budget per analysis chunk
Force chunk break after this many days
Auto-fail stuck runs after this many minutes
Task Monitor

No active tasks

Hosted MCP server
Let supported AI clients (Claude Desktop, Cursor) query this workspace through RelationTrace's hosted Model Context Protocol endpoint. OAuth-secured; workspace-scoped read access only.
Connection details
Most clients auto-discover authorization from this URL. Paste it into the MCP server config.
—
Connect a client (instructions)
  1. In Claude Desktop, open Settings → Developer → MCP servers.
  2. Add a new server with type Streamable HTTP and the URL above.
  3. Approve the OAuth consent screen that opens in your browser. Pick this workspace.
  4. Return to this page; the connected client and active session will appear below.
Connected clients

Loading…

Active sessions

Loading…

Usage
Tool calls counted against this workspace's monthly quota.
Personal access tokens
Long-lived bearer tokens for CLI tools that don't expose an OAuth scope knob. Pick the scopes you need at mint time; the token is shown once.

Loading…

Imported Files
Files currently in the workspace.

Loading...

Re-run Operations
Re-process data without re-importing. Useful after model or parameter changes.
Danger Zone
Destructive operations. These cannot be undone.
Removes signals, findings, evidence, phases. Keeps messages.
Removes everything. Start fresh.

Suggested Duplicates

Workspace
Manage workspace access and invitations.
Waiting List
Review and manage waiting list applications.
User Management
View and manage all registered users.

Select a prompt to view.