Life Graph Engine
The intelligence layer behind Anima.
Every voice entry becomes a node in your personal graph. Over time, patterns emerge that no human could spot manually.
What Is the Life Graph?
When you talk about your day, Anima does not just classify activities into stats. It builds a personal knowledge graph connecting people, places, activities, emotions, and time. Every entry adds new nodes and new connections, creating a rich, evolving model of your life.
The Five Layers
Each layer builds on the one before it. By the time you reach Layer 5, the system understands your life well enough to hold meaningful conversations about it.
You talk. AI listens and classifies activities into seven stats: Strength, Vitality, Intellect, Empathy, EQ, Creativity, and Awareness.
Activities become nodes. Nodes connect to people, places, emotions, and time. Your personal graph grows with every entry.
After 2-3 weeks, the graph reveals patterns: "Your creativity spikes 3x on days you exercise." These are insights only the graph can surface.
Seven AI agents, one per stat, that know your personal history and can have conversations with you. Your Strength agent knows every workout. Your Creativity agent knows every project.
All seven agents deliberate on your life questions from different dimensional perspectives. Ask a hard question, and get seven angles on the answer, grounded in your own data.
Pattern Examples
These are the kinds of patterns Anima detects once your graph has enough data. They are specific to you, grounded in your own entries, and impossible to spot through willpower alone.
"Your Creativity peaks on days you also score high in Vitality."
"You mention 'tired' 3x more on weeks with zero Strength activity."
"Conversations with [Sarah] correlate with +40% EQ scores the following day."
"Your Awareness score drops when Intellect exceeds 60% of total XP for the week."
"Your best weeks happen when no single stat dominates. Balance is your superpower."
How It Grows Over Time
Your Life Graph starts simple and becomes increasingly powerful. Here is what unlocks at each stage.
Under the Hood
Each user has their own isolated knowledge graph. Nodes represent entities (activities, people, places, emotions, time slots) and edges represent observed relationships between them. The graph is append-only and temporal, meaning every edge carries a timestamp so patterns can be analysed across time windows.
When a stat agent needs context for a conversation, it does not receive your entire graph. Instead, a relevant sub-graph is serialised and injected into the agent's context window. This keeps conversations focused and token-efficient while still grounding them in your real data.
Each of the seven stat agents maintains a compressed memory of your history within its domain. This includes key milestones, recurring patterns, and notable changes in behaviour. The agent can reference specific entries and draw on longitudinal trends when responding to your questions.
The pattern engine runs sliding-window analyses across your graph, looking for statistically significant correlations between node types. It accounts for day-of-week effects, seasonal trends, and lag correlations (where one activity influences a stat hours or days later). Only patterns that cross a confidence threshold are surfaced to you.