↑ ↓ or scroll

The Agentic Memory Landscape

Memory, Context
& Graphs

How agent memory, context engineering, and knowledge graphs are converging in 2026, and where Living Memory and Alexandria fit.

The Problem

An army of amnesiacs.

Agents with skills and tool access can do remarkable work. But without shared memory, every session starts from zero. No lessons from last time. No precedent. No pattern recognition across sessions or across agents.

Without memory, agency does the same work repeatedly. An agent fleet without shared memory is an army of amnesiacs.

Memory is the multiplier; agency is the force. The question is: what kind of memory, and how does it compound?

The Evolution

From retrieval to learning.

2023
Naive RAG
Chunk documents, embed, retrieve by similarity. Read-only, one-shot. The agent gets "related text," not understanding.
2024
Advanced RAG
Hybrid search, reranking, HyDE, corrective loops. Better retrieval, but still stateless. The system doesn't learn from use.
2025
Agentic RAG
The agent decides what to retrieve, when, and how deep. Multi-hop reasoning. Adaptive routing. Still read-only at its core.
2026
AI Memory
Read-write. The system learns from every session. Shared across agents. Knowledge compounds, decays, self-maintains.

Each step addresses a limitation of the last. RAG gave agents knowledge. Agentic RAG gave them judgment about retrieval. AI Memory gives them the ability to learn, remember, and forget.

The 2026 Shift

Context engineering, not prompt engineering.

The industry consensus in 2026: the bottleneck isn't the model, it's the context the model receives. "Context engineering" is the discipline of assembling the right knowledge, at the right moment, at the right depth, for the right agent.

This is more than prompt design. It spans retrieval, filtering, scoping, progressive disclosure, token budgets, and deciding what to exclude. The model is only as good as what you put in the window.

Our entire architecture is a context engineering system. getContext, retrieval filters, workflow-scoped precedent, token budgets, progressive disclosure: these are all context engineering.

Prompt engineering
Craft the instruction. Static. One-size-fits-all.
RAG
Retrieve relevant chunks. Better, but blunt: similarity isn't understanding.
Context engineering
Assemble the full picture: lessons, precedent, entity context, scoped by workflow, bounded by token budget, with judgment about what to include and what to leave out.

Why Graphs

Search finds what's near your query.
Graphs find what's near your understanding.

Every sophisticated RAG pattern either adds graph structure (Graph RAG, Hybrid RAG) or adds agentic reasoning to compensate for search limitations (Adaptive RAG, Agentic RAG). Each escalation is a workaround for something curated graph traversal handles natively.

Graphs provide two things embeddings can't: typed relationships (decisions link to assumptions, assumptions link to evidence) and structural discovery (the third hop takes you to territory you didn't aim for). Search can't escape the query neighborhood. Graphs can.

Vector search Knowledge graph
Finds Text similar to query Entities connected to understanding
Discovery Locked to query neighborhood Multi-hop traversal across topics
Relationships Implicit (cosine distance) Explicit, typed, directional
Maintenance Re-embed on change Update claims, prune, challenge
Gives the agent Related text Reasoning structure

Context Graphs

The enterprise opportunity.

Foundation Capital calls context graphs "AI's trillion-dollar opportunity." The argument: systems of record capture what happened. Context graphs capture why, who, when, and what precedent. Decision traces, not just state changes.

Salesforce knows the opportunity stage moved. It doesn't know who approved the deviation, what precedent was referenced, or what the state was when the decision was made.

The gap isn't missing data. It's missing decision traces. Agents run into the same ambiguity humans resolve with judgment and organizational memory. But the inputs to those judgments aren't stored as durable artifacts. Context graphs make them durable, queryable, and shared.

Five Levels

From static files to organizational memory.

1
Basic Context
Skills, rules, scratchpads. No learning loop. Every session starts the same.
2
Reflections
Session-end lessons stored in a DB. getContext at start. Light pruning.
3
Living Memory
Layered scope, entity anchoring, forgetting by design, multi-workflow.
4
Agentic Memory
Shared learning loop. Memory manages itself. Feeds and draws from the store.
We are here
5
Alexandria
Org-wide library. Knowledge graph. Propositional claims. Progressive disclosure.
Vision

Most teams are at Level 1 or 2. We operate at Level 4 today: agents learn from each other's sessions, share precedent across workflows, and the memory system maintains itself. Level 5 is the full org-wide knowledge library.

Architecture

Memory. Learning. Personalization.

The architecture decomposes into three layers running on different timescales. Most systems have Memory. Few have Learning. Almost none have all three.

Personalization
Right knowledge, right agent, right moment (getContext, retrieval filters, token budgets)
Real-time
Learning
Raw experience becomes structured knowledge (reflection, digestive pipeline, condensation, forgetting)
Async
Memory
Where knowledge lives (lessons DB, scratchpads, knowledge graph, Alexandria claims)
Storage

Why the seams matter

A Memory failure is data loss. A Learning failure is a missed insight or false lesson. A Personalization failure is serving the wrong context or overwhelming the agent. Naming the layers prevents conflating a retrieval problem with a storage problem.

Agent memory vs. agentic memory

Agent memory gives agents access to a store. The store is passive. Agentic memory manages itself via tool calls: ingesting, evaluating, connecting, consolidating, and forgetting. The memory layer is not a database with an API. It is an active participant in its own maintenance. We are building agentic memory.

What's Different

Not storage. Metabolism.

Capability Typical RAG / vector store Living Memory + Alexandria
Retrieval Embedding similarity, one-shot Workflow-scoped, entity-anchored, multi-pass (berrypicking), token-budgeted
Learning None. Re-embed on manual update. Session-end reflection, digestive pipeline, lessons shared across agents
Forgetting None. Accumulate forever. Decay tiers, condensation, supersession, stale detection. Forgetting is architecture.
Connections Implicit (cosine distance) Typed, weighted, entity-anchored. Decisions link to assumptions link to evidence.
Quality Whatever was embedded Trust tiers, immune system (write-time validation), quality-gated claims
Presentation Chunks returned by query Claims installed as capabilities. [[wiki-links]] carry reasoning inline. Progressive disclosure.

Pressure-Tested

Validated against 18 external sources.

We pressure-tested the architecture against LangChain, Engram, ALMA, MemGPT/Letta, Cognee, MAPLE, Anthropic's memory tool, a major 2026 academic survey, 8 RAG architecture patterns, and more. What we found:

Validated strengths

Hierarchical token-level memory (academic gold standard). Memory evolution (decay, condensation, supersession) ahead of the field. Propositional framing matches the "retrieve-then-generate" paradigm the industry is moving toward. Forgetting design is the most complete in any source evaluated.

Gaps we're closing

Episodic memory (raw case records beneath lessons). Context pressure handling for long-running agents. Shared memory access control for multi-agent writes. Memory explainability and retrieval attribution. Event-triggered learning for critical failures.

Borrowed from the field

Decay tiers (Engram). Memory strengthening (Cognee). Offline consolidation cycle (CLS theory). Context pressure events (Anthropic). Reflect-generate-verify for methodology changes (ALMA). M/L/P naming (MAPLE).

Core Differentiators

Where we lead.

Claims as installed capabilities
Loading a claim doesn't retrieve a record. It gives the agent reasoning it didn't have before. A vague claim is a buggy function. A stale claim is a regression.
Forgetting is first-class
Decay tiers, condensation, supersession, archival. Not a cleanup job. A structural property. The knowledge system that can't prune is the brain that can't think.
Methodology over tooling
Teams don't change their tools. They publish claims in a standard format and query context through a standard interface. The methodology is the coordination surface.
Encountered, not just sought
High-value knowledge is encountered during reading via [[wiki-links]]. Agents encounter reasoning inline, not just query results. Search gives you content. Graphs give you structure.
Shared across people and agents
Every correction, every lesson, every connection improves the graph for the next person and the next agent. The loop compounds across the organization.
Graph-first architecture
The industry starts from Naive RAG and patches toward graphs. We start from the graph. Curated traversal handles natively what RAG escalations work around.

The Landscape

The industry is converging
on what we're building.

RAG is adding graphs. Graphs are adding agents. Agent memory is adding learning loops and forgetting. Context engineering is becoming the discipline that ties them together. Living Memory and Alexandria sit at the intersection: agentic memory on a knowledge graph, with context engineering as the serving layer.

Memory is the multiplier. Agency is the force.