AutoGen Memory
https://microsoft.github.io/autogen/
ListMemory chronological context + teachable agents that vectorise corrections. Integrates with Mem0/Zep rather than building deep memory natively.
At a glance
- Type
- List + teachable vectors
- Tier
- T2
- Section
- Framework-embedded memory
- Created
- 2023-08
- Latest release
- python-v0.7.5 2025-09-30
- License
- MIT
- GitHub
- 57.7k★ +1.7k/mo Python
- Pricing
- OSS (MIT); no direct cost for library; Azure AI Foundry Agent Service for managed hosting uses Azure consumption pr…
- Funding
- Microsoft-backed; AutoGen is an open-source Microsoft Research project with no independent funding; merged into Mic…
Taxonomy
- storage
- kv
- retrieval
- injection
- persistence
- cross-session
- update
- append-only
- unit
- episode
- governance
- inspectable
- conflict
- append
When to use
Optimised for: Microsoft Research multi-agent + teachable corrections
Anti-fit: not for non-AutoGen stacks
Pros & cons
Pros
Microsoft Research origin; strong multi-agent conversation memory model.
Cons
AutoGen's broader feature drift over time has affected memory abstractions; less stable than dedicated layers.
Claims & capabilities
Framework adoption nearly doubled YoY: ~9% of organizations early-2025 to ~18% early-2026 per Datadog State of AI Engineering. AutoGen now in maintenance mode — superseded by Microsoft Agent Framework. AutoGen Bench provides agent benchmark suite; matches LangGraph in token use and latency
Technical surface
- API surface
- SDK: Python, .NET
- Backend storage
- pluggable
- Deployment
- Both (OSS self-hosted Python; Azure AI Foundry managed runtime with GA Q1 2026)
- Embedding model
- multiple supported
- Multi-tenancy
- namespace
- MCP
- via official adapter — autogen-ext-mcp
- A2A
- searched not found
- OpenTelemetry
- first-class — AutoGen has OTel runtime
Similar systems
Other framework-embedded memory in the catalog, ranked by inbound references.
- LangGraph Persistence T2
Distinct from LangMem. Built-in checkpointer saves graph state per super-step (short-term, thread-scoped). Store System adds long-term hierarchical key-value memory across threads with optional vector search + TTL. Postgres / Mongo / Redis stores for production.
- CrewAI Memory T2
Memory subsystem inside the CrewAI orchestration framework; integrates with Mem0 for the long-term tier.
- AGiXT Adaptive Memory T2
Open-source AI automation platform. Routes between short-term and long-term memory adaptively across any LLM provider; plugin system for storage backends. Memory managed at the instruction-management layer — task context, instruction state, conversation history as unified agent state.
- Agno (Phidata) Memory T2
Agno (formerly Phidata). AgentStorage persists sessions to a DB; AgentMemory auto-classifies/store user preferences and conversation summaries. Single-line integrations with LanceDB, Pinecone, Weaviate, Qdrant.
- Botpress LLMz T1
Per-plan vector-DB storage quota + LLMz autonomous engine (in-session working memory) + Knowledge Base (semantic search over uploaded docs). Long-term user memory persists across sessions.
- DSPy History T3
dspy.History primitive — typed field holding messages: list[dict] that slots into any Signature . No persistent memory of its own; purely a structured context-injection contract. DSPy's optimisation loop (MIPRO, BootstrapFewShot) treats historical turns as trainable few-shot structure.
Related systems
References (2)
- Mem0 integrates with — ListMemory chronological context + teachable agents that vectorise corrections. Integrates with Mem0/Zep rather than building deep memory natively.
- Zep & Graphiti integrates with — Integrates with Mem0/Zep rather than building deep memory natively.
Referenced by (1)
- mcp-memory-service (doobidoo) integrates with — Agent-pipeline focused (LangGraph, CrewAI, AutoGen)