What can you build with real memory?

31 ways individuals, teams, and organizations use Rekall — across personal memory, team hives, production agents, and persistent husk agents.

account_tree

Graph-backed, not just a vector store

Entities and relationships live in a real knowledge graph (Neo4j), every extracted fact carries dates, and decisions are structured bi-temporal records — alternatives, rationale, outcome, supersession. The measured payoff is point-in-time recall: +14.1pp on temporal questions, with correct refusals improving too. Lots of teams are chasing this; Rekall measures it.

all_inclusive

Infinite context, finite windows

Progressive disclosure returns a compact index first and expands only what the task needs — months of memory stay accessible without ever flooding a prompt. Unlimited memory is useless if it blows the context window; Rekall is built so it never does.

join_full

These aren’t modes — run them all

Every use case below runs on the same memory system, at the same time. Scopes keep personal, team, and agent memory isolated — measured at zero cross-tenant leaks across 23,840 checked retrievals, with team-scoped recall matching the single-tenant control. Each one makes the next better.

emoji_objects

These are starting points, not a menu

Everything above is built from the same primitives — memory layers, hives, a knowledge graph, agents, gates, and webhooks — composed differently. They're suggestions to get you started. If you can imagine a way to use persistent, graph-backed memory, you can build it on Rekall.

Grab the primitives and build your ownarrow_forward

Which one is yours?

Every use case starts the same way: connect a tool, and Rekall starts remembering. Free for personal use — no credit card required.