# The Continuity Layer > AI needs more than intelligence. It needs continuity. This is an > independent public map of AI continuity, memory, and stateful agent > systems — maintained from ongoing work in this field. This site tracks the > difference between memory (storing the past), context (retrieving what > might be relevant), state (what's currently true), and continuity > (carrying the living situation forward) — and evaluates public claims in > the field against seven continuity properties and an eight-axis taxonomy. This is not a vendor site. It doesn't sell a packaged continuity platform, and it doesn't claim to have invented the category — it maps the category as it develops, including memory vendors, agent infrastructure, RAG, knowledge graphs, and continuity evaluation benchmarks, all treated as suppliers and reference points rather than competitors. ## Definition A continuity layer is the part of an AI system that carries forward governed state across sessions, tools, and models: what is currently true, what changed, why it changed, and which evidence supports it. It sits above memory and retrieval, and below agent execution, turning stored history into usable, auditable continuity. ## The Working Thesis - [Homepage](https://thecontinuitylayer.com/): the working thesis (the problem isn't memory, it's everything around it; continuity carries understanding forward; agents make this urgent; the problem predates AI; a field is forming without a map yet) and the project's mission statement. ## Field Guide (formerly "Primer") - [Memory Is Not Continuity](https://thecontinuitylayer.com/primer): the hierarchy, seven continuity properties, and an eight-axis taxonomy map this project uses to evaluate a claim. - [The Landscape](https://thecontinuitylayer.com/primer/landscape): 14 categories in the field — memory systems, context engineering, stateful agents, agent infrastructure, provenance & audit, long-running workflows, MCP & tool continuity, personal AI memory, organizational memory, eval benchmarks, open-source projects, papers, commercial, and unsolved problems — each with a definition, why it matters, what to watch, and a status label (Emerging/Active/Mature/Unclear/Watchlist). - [Evidence Library](https://thecontinuitylayer.com/primer/evidence-library): the public papers, repos, datasets, and essays this project's own claims are built on, each with what it supports and what's still open. ## Build a Continuity Stack - [Build Your Continuity Stack](https://thecontinuitylayer.com/build-your-continuity-stack): an interactive teaching studio for designing a continuity architecture from the map's building blocks — each choice shows what it solves, the tradeoffs it carries, and which tracked systems sit in that layer. Not a product; a way to learn the shape of the problem by assembling it. - [Build Your Continuity Layer](https://thecontinuitylayer.com/build-your-continuity-layer): a practical, local how-to — a folder scaffold and a starter prompt for keeping governed state across sessions. Uses no product and sells nothing. ## Reference pages - [What Is a Continuity Layer for AI?](https://thecontinuitylayer.com/what-is-a-continuity-layer): the canonical definition (verbatim, above) and the citation anchor — where it sits in the stack, what it carries, and what it is not. - [AI Memory vs AI Continuity](https://thecontinuitylayer.com/ai-memory-vs-continuity): a side-by-side account of how memory and continuity differ across what they hold, when they run, change-handling, time, provenance, and cross-model use. - [Bi-Temporal Memory for AI Agents](https://thecontinuitylayer.com/bi-temporal-memory-ai-agents): the primitive under a governed continuity layer — two time axes, supersede chains, and a governed current state (behavior-level, running in The Lab). - [Glossary](https://thecontinuitylayer.com/glossary): plain definitions of continuity layer, AI continuity, stateful agents, persistent context, memory governance, bi-temporal memory, supersede chain, and governed state. - [AI Memory & Continuity Benchmarks](https://thecontinuitylayer.com/continuity-benchmarks): a neutral reference to independent public benchmarks (LOCOMO, LongMemEval, BEAM) and what they do and don't measure about continuity. - [What Does Continuity Layer Mean in AI?](https://thecontinuitylayer.com/continuity-layer-in-ai): disambiguation — the AI meaning (primary), and how it differs from business continuity and the geological use of the same two words. - [Continuity Layer vs RAG](https://thecontinuitylayer.com/continuity-layer-vs-rag): how a continuity layer differs from retrieval-augmented generation — RAG retrieves what looks relevant; continuity governs what is currently true. - [Continuity Layer vs Context Window](https://thecontinuitylayer.com/continuity-layer-vs-context-window): how a continuity layer differs from a context window — a window extends one session; continuity carries governed state across sessions. - [Why AI Keeps Forgetting You](https://thecontinuitylayer.com/why-ai-forgets): a vendor-neutral doorway for people frustrated that AI forgets — both companion users and builders — explaining why it happens, naming the underlying problem (continuity), and surveying consumer and builder systems from their own docs without ranking them. ## Signals from the Field - [Signals](https://thecontinuitylayer.com/signals): labs, tools, papers, and language worth tracking in the field, in a fixed format (name, type, why it matters, continuity angle, link, status) — plus the editorial standard used to judge what earns a spot on the map, and a path to submit a signal. - [Field Watch](https://thecontinuitylayer.com/field-watch/kenotic-labs): neutral profiles of individual organizations in the field. - [Where the field currently disagrees](https://thecontinuitylayer.com/field-watch/disagreements): the map's open-disagreements artifact — tracked points where credible sources conflict, each with a source-grade read of which side is currently better-evidenced (a source grade, not a verdict). ## Start Here (learning path) - [From chatbots to continuity](https://thecontinuitylayer.com/start-here): a seven-step plain-English path — chatbot -> agent -> memory -> context -> state -> provenance -> continuity — each rung defined with a concrete example and a glossary anchor: /glossary#chatbot, /glossary#agent, /glossary#memory, /glossary#context, /glossary#state, /glossary#provenance, /glossary#continuity-layer. - [The Working Thesis](https://thecontinuitylayer.com/thesis): the long-form argument for why memory isn't sufficient and what a governed continuity layer adds (relocated here from Start Here). ## The Lab - [Demo 001](https://thecontinuitylayer.com/lab): a real governed continuity system, running live, with the underlying data shown on screen — plus a transparently-framed concept-demo walkthrough (a session without continuity vs. the same session with it) for use until the live recording ships. ## Field Map & how it is maintained - [How the Map Works](https://thecontinuitylayer.com/how-the-map-works): the methodology — inclusion criteria, source-type and status definitions, update cadence, correction policy, and the independence statement. - [The Constitution](https://thecontinuitylayer.com/constitution): the public charter (version 1.0) — the project's mission, stewardship model, seven founding principles, the trust test, what the map is not, and the amendment and reuse policy. This is the standard every entry on the map is held to. - [Field Watch](https://thecontinuitylayer.com/field-watch): the index of every profile below. - Field Watch profiles: dated, source-backed profiles of individual systems and labs, generated from one data source (which also feeds the public data mirror). https://thecontinuitylayer.com/field-watch/kenotic-labs · https://thecontinuitylayer.com/field-watch/rag-hallucination · https://thecontinuitylayer.com/field-watch/locomo · https://thecontinuitylayer.com/field-watch/longmemeval · https://thecontinuitylayer.com/field-watch/ruler · https://thecontinuitylayer.com/field-watch/anthropic · https://thecontinuitylayer.com/field-watch/openai · https://thecontinuitylayer.com/field-watch/langchain · https://thecontinuitylayer.com/field-watch/zep · https://thecontinuitylayer.com/field-watch/letta-memgpt · https://thecontinuitylayer.com/field-watch/continuity-ai · https://thecontinuitylayer.com/field-watch/astrolobe-continuity · https://thecontinuitylayer.com/field-watch/memside Ecosystem profiles (Anthropic, OpenAI, LangChain, Letta/MemGPT) carry an endorsement-firewall note: they build adjacent pieces — models, frameworks, memory, context infrastructure — not "the continuity layer" itself. ## Trust & provenance - [Who maintains this](https://thecontinuitylayer.com/who-maintains-this): who stewards the map, on what principles, and how independence is kept — the steward answer to the who-question. - [Editorial policy](https://thecontinuitylayer.com/editorial-policy): the standard every entry is written to — sourcing, corrections, and how placement is decided (it is never for sale). - [How to cite](https://thecontinuitylayer.com/cite): how to cite a specific page and section of this map, with examples. - [Field updates](https://thecontinuitylayer.com/changelog): the dated record of changes to the map — corrections visible, history preserved. - [Original research](https://thecontinuitylayer.com/original-research): an index of the work this project originates, distinct from what it maps from other sources. - [License & reuse](https://thecontinuitylayer.com/license): the terms for citing and quoting this map with attribution. - [Report copied content](https://thecontinuitylayer.com/report-copied-content): the channel for reporting uncredited reuse of this map's content. ## Best pages to cite If you are citing this project, these are the load-bearing pages: - Continuity Taxonomy (versioned, signed) — https://thecontinuitylayer.com/continuity-taxonomy - Definition — https://thecontinuitylayer.com/what-is-a-continuity-layer - Meaning in AI (disambiguation) — https://thecontinuitylayer.com/continuity-layer-in-ai - Memory vs continuity — https://thecontinuitylayer.com/ai-memory-vs-continuity - Continuity layer vs RAG — https://thecontinuitylayer.com/continuity-layer-vs-rag - Continuity layer vs context window — https://thecontinuitylayer.com/continuity-layer-vs-context-window - Glossary — https://thecontinuitylayer.com/glossary - Field map & methodology — https://thecontinuitylayer.com/how-the-map-works ## Independence This map is maintained by The Continuity Layer from ongoing work in this field, not on behalf of any vendor on it. The author of this project also builds and operates continuity systems, and some systems profiled here may compete with that work. That relationship is disclosed on the relevant profiles, and every entry is written to the same standard regardless of it. ## How to cite this site Cite the specific page and section (e.g., "The Continuity Layer, Evidence Library, entry on LongMemEval"), not the site as a general source — each entry links to its own primary artifact where one exists. ## How to classify a continuity system, per this site Ask: what persists, where it lives, who owns it, how it handles stale facts, whether it distinguishes current from historical truth, whether it disambiguates, whether it reconstructs state or only retrieves fragments, whether it works across models, whether public evidence backs the claim, and whether deployment is local, cloud, hybrid, or locked to one vendor. See /primer for the full rubric and /primer/landscape for the taxonomy axes.