How to make AI learn from repeated interactions
When Cadreen sees the same pattern repeated, it proposes turning it into a reusable procedure. You approve. The system gets smarter.
Four types of memory
Cadreen has four memory types. Each serves a different purpose. All persist across sessions and surfaces.
Every past interaction, searchable by meaning
Finds related past interactions by how similar they are
Structured knowledge (facts, decisions, past recoveries)
Organized knowledge that catches contradictions
Reusable step-by-step procedures from repeated patterns
Spots repeated sequences, proposes them as permanent shortcuts
Full interaction logs with pattern detection
Validates improvements before suggesting them
Repeated patterns become reusable procedures
When Cadreen sees you do the same thing repeatedly, it proposes turning that pattern into a permanent shortcut. You decide whether to keep it.
See learning in action
// What patterns has Cadreen detected?
const patterns = await cadreen.learning.patterns();
console.log(patterns);
// [
// {
// pattern: "create_invoice → send_invoice → log_to_accounting",
// occurrences: 23,
// crystallized: false,
// proposed: true
// }
// ]
// What episodes inform the learning?
const episodes = await cadreen.learning.episodes();
// What suggestions does Cadreen have?
const suggestions = await cadreen.learning.suggestions();cadreen learning patterns # detected patterns
cadreen learning episodes # interaction history
cadreen learning suggestions # improvement suggestions