Dedicated memory layers

German enterprise stack. PhariaCatch captures + preserves organisational knowledge through structured human-machine collaboration; PhariaAssistant queries via semantic search. Combines sovereign deployment constraints with explainability…

"Extract–Cognify–Load" pipeline that turns raw input into a typed, queryable knowledge graph for agent recall.

Standalone memory service from Vectorize. Open source. Biomimetic four-network design (World, Bank, Observation, Opinion). Ships an MCP memory server.

German startup. Captures how real work is performed across enterprise systems and teams; converts process knowledge into a persistent memory layer agents use to automate complex workflows. Targets the tacit-knowledge gap that agents curr…

Israeli production model. Interleaves Mamba SSM layers with Transformer attention (1:7 ratio) + MoE blocks every 2 layers. SSM layers maintain fixed-size recurrent hidden state — compressed long-context memory bypassing growing KV cache.

LLM-as-Operating-System paradigm — model manages its own memory, paging between core context, recall buffer, and archival store. UC Berkeley origin; transitioning to letta_v1_agent .

Universal memory layer for AI agents. Three concurrent stores (vector + graph + KV); LLM-extracted facts; concurrent retrieval via ThreadPoolExecutor.

Reasoning-focused successor to HyperCLOVA X. Pre-trained on ~6T high-quality Korean + English tokens. 128K context via rotary position embeddings + grouped-query attention. Sovereign reasoning LLM for Korean enterprises and government.

Builds agents that learn from experience by constructing a "world model" of the micro-environment they operate in. Specialise over time through Agentic RL loop rather than relying on static prompt context. Targets the ~50% task-completio…

Cloud-free agent memory using information-geometric retrieval (Fisher-Rao distance + BM25 + entity graph + temporal fusion via 4-channel RRF) instead of LLM calls for scoring and lifecycle. Mode A is the only system to cross 70% LoCoMo w…

Memory engine with API, app, browser extension, and MCP server. Extracts facts, tracks updates, resolves contradictions, auto-forgets expired info. Plugins for Claude Code, OpenCode, OpenClaw, Hermes.

Bi-temporal knowledge graph (event time + ingestion time). Strong on chronological reasoning and contradiction tracking. Graphiti is the open-source core.

SEA-LION-Embedding (March 2026): retrieval + reranking models contrastively trained on 245M text pairs across 10 SE Asian languages. SEA-BED benchmark (169 datasets). SEA-LION v4 (Gemma-based) at 128K context with native function calling.

Native graph-based memory for autonomous AI agents. Bio-inspired sleep-based memory consolidation.

Tsinghua spinout. Foundation agent series for autonomous smartphone/PC GUI control. Persistent task memory retains shopping history, app usage patterns, multi-step task state across sessions. AutoGLM-Thinking claimed first agent capable …

Agent memory product evaluated against LoCoMo.

Agent-memory product. LLM-curated hierarchical context.

Korean startup. Connects scattered internal/external data across multiple chats and services to improve AI answer quality at retrieval. 18-year-old CEO Jang Jun-ha; pitch is that answer quality depends on what the model remembers, not wh…

Opensource memory layer focused on programming agents — works with Cursor, Windsurf, Claude Code, Roo Code via MCP. Maintains a coding-context memory across sessions and across IDEs.

Connects Slack, Gmail, Notion, Drive so agents can recall and reason across company knowledge.

India's first agentic AI assistant. Persistent memory of prior interactions across sessions; accepts voice/text/image/document input building continuous context. Decomposes goals into sub-tasks executed by parallel/sequential agent chains.

ByteDance Seed + Zhejiang Univ + SJTU. Multimodal agent with long-term memory as entity-centric graph built from live video + audio streams. Parallel memorisation + control processes. Ships M3-Bench evaluation dataset.

Open-source long-term memory framework for autonomous agents.

Multi-layered open-source memory system for AI agents.

User-profile-based long-term memory. Per-user profile + event timeline. API + Python/Node/Go SDKs + MCP server. Profile-centric rather than fact-centric.

Shenzhen. Argues attaching external vector stores is architecturally flawed; trains memory as an intrinsic capability via Agentic RL. Plans B2B enterprise KM and C2C personalisation in 12 months. Founded by Tianjin University AI professor.

Structured, persistent memory layer for AI agents. Open source.

Memory layer from MatrixOrigin. Strong-retrieval / clear-reader-separation design.

YC P26 (Spring 2026). Unified memory layer so all agents on a team share one persistent context — meetings, Claude sessions, Slack threads all become queryable. Solves the "goldfish brain" problem of every session starting from zero.

Single-file memory layer (one .mv2 file). No DB, no server. Append-only sequence of immutable Smart Frames with timestamps + checksums. Native Rust core (rewritten from Python).

Walrus is a Sui-blockchain decentralised storage protocol; MemWal is its AI-facing SDK that stores encrypted agent memory on-chain — portable, verifiable, shareable across agent systems with no single-vendor lock-in. Targets production a…

AI-native storage tier for gigascale agentic inference. Bridges GPU HBM and general-purpose storage with petabyte-scale RDMA-accelerated KV-cache placement; coordinated via DOCA, Spectrum-X Ethernet, NIXL library, Dynamo software. Aligne…

Deploys agents across the web to build structured profiles of individuals from public digital footprint, then feeds the people graph into AI agents as contextual memory so they understand users before any conversation starts.

Omni memory system for personalised, long-horizon, self-evolving agents. From OPPO PersonalAI.

Memory engine product. Pitch is high-accuracy retrieval at very low latency.

Open-source cognitive memory engine for LLMs and agents.

Long-term memory engine for developers — captures everything the user does (browser, IDE, terminal) on-device, exposes via MCP for any IDE/agent.

YC S25 + $3M seed (Feb 2026). Workflow orchestration mapping corporate environments as a structured context graph so agents always have institutional knowledge for tasks. CTO frames differentiator as "context engineering."

Node.js/TypeScript-native memory layer. SQLite + local Transformers.js embeddings (~80MB model via WebAssembly). AUDN loop (Add/Update/Delete/None) resolves contradictions before write rather than at retrieval — read-after-write consiste…

SQL-native memory engine for AI agents — single library, opt-in conscious/auto modes, structured Pydantic outputs. Aims to be 'one line of code' to add memory to any LLM workflow.

Memory layer for AI assistants — semantic short-term and long-term memory with concept-graph linking and decay/reinforcement.