mengram-ai 2.11.0

Human-like memory for AI — semantic, episodic & procedural. Experience-driven procedures, Cognitive Profile, unified search, memory agents. Free open-source Mem0 alternative.

Mengram

The memory layer for AI agents that learns from experience

Your agents remember facts, events, and workflows — and procedures improve automatically when they fail.

Website · Get API Key · API Docs · Examples

---

Why Mengram?

Every AI memory tool stores facts. Mengram stores 3 types — and procedures evolve from failures.

	Mengram	Mem0	Letta	Zep
Semantic Memory (facts)	✅	✅	✅	✅
Episodic Memory (events)	✅	❌	Partial	❌
Procedural Memory (workflows)	✅	❌	❌	❌
Experience-Driven Evolution	✅	❌	❌	❌
Cognitive Profile	✅	❌	❌	❌
Knowledge Graph	✅	✅	✅	✅
LangChain / CrewAI	✅	Partial	❌	✅
Import (ChatGPT, Obsidian)	✅	❌	❌	❌
MCP Server	✅	✅	✅	❌
Price	Free	$19–249/mo	Free (self-host)	Enterprise

Quick Start

pip install mengram-ai

from cloud.client import CloudMemory

m = CloudMemory(api_key="om-...")  # Free key → mengram.io/dashboard

# Add a conversation — Mengram auto-extracts facts, events, and workflows
m.add([
    {"role": "user", "content": "Deployed to Railway today. Build passed but forgot migrations — DB crashed. Fixed by adding a pre-deploy check."},
])

# Search facts
m.search("deployment setup")

# Search events — what happened?
m.episodes(query="deployment")
# → [{summary: "Deployed to Railway, DB crashed due to missing migrations", outcome: "resolved", ...}]

# Search workflows — how to do it?
m.procedures(query="deploy")
# → [{name: "Deploy to Railway", steps: ["build", "run migrations", "push", "verify"], ...}]

# Unified search — all 3 types at once
m.search_all("deployment issues")
# → {semantic: [...], episodic: [...], procedural: [...]}

JavaScript / TypeScript:

npm install mengram-ai

const { MengramClient } = require('mengram-ai');
const m = new MengramClient('om-...');

await m.add([{ role: 'user', content: 'Fixed OOM with Redis cache' }]);
const all = await m.searchAll('database issues');
// → { semantic: [...], episodic: [...], procedural: [...] }

Experience-Driven Procedures

The feature no one else has. Procedures learn from real outcomes — not static runbooks.

Week 1:  "Deploy" → build → push → deploy
                                         ↓ FAILURE: forgot migrations, DB crashed
Week 2:  "Deploy" v2 → build → run migrations → push → deploy
                                                          ↓ FAILURE: OOM on Railway
Week 3:  "Deploy" v3 → build → run migrations → check memory → push → deploy ✅

This happens automatically when you report failures:

# Report failure with context → procedure evolves to a new version
m.procedure_feedback(proc_id, success=False,
                     context="OOM error on step 3", failed_at_step=3)

# View version history
history = m.procedure_history(proc_id)
# → {versions: [v1, v2, v3], evolution_log: [{change: "step_added", reason: "prevent OOM"}]}

Or fully automatic — add conversations and Mengram detects failures, links them to procedures, and evolves:

m.add([{"role": "user", "content": "Deploy to Railway failed again — OOM on the build step"}])
# → Episode auto-linked to "Deploy" procedure → failure detected → v3 created

Cognitive Profile

One API call generates a system prompt from all your memories:

profile = m.get_profile()
# → "You are talking to Ali, a developer in Almaty building Mengram.
#    He uses Python, PostgreSQL, and Railway. Recently debugged pgvector deployment.
#    Workflows: deploys via build→twine→npm→git. Communicate directly, focus on practical next steps."

Insert into any LLM's system prompt for instant personalization.

Import Existing Data

Kill the cold-start problem — import your ChatGPT history, Obsidian vault, or text files:

# ChatGPT export (Settings → Data Controls → Export)
mengram import chatgpt ~/Downloads/chatgpt-export.zip --cloud

# Obsidian vault
mengram import obsidian ~/Documents/MyVault --cloud

# Any text/markdown files
mengram import files notes/*.md --cloud

Works with Python SDK too:

m = CloudMemory(api_key="om-...")
m.import_chatgpt("export.zip")
m.import_obsidian("~/Documents/MyVault")
m.import_files(["notes.md", "journal.txt"])

Integrations

MCP Server (Claude Desktop, Cursor, Windsurf)

{
  "mcpServers": {
    "mengram": {
      "command": "mengram",
      "args": ["server", "--cloud"],
      "env": { "MENGRAM_API_KEY": "om-..." }
    }
  }
}

LangChain

from integrations.langchain import MengramChatMessageHistory, MengramRetriever

# Drop-in message history — auto-saves to Mengram
history = MengramChatMessageHistory(api_key="om-...", session_id="session-1")

# RAG retriever — searches all 3 memory types
retriever = MengramRetriever(api_key="om-...")

CrewAI

from integrations.crewai import create_mengram_tools

tools = create_mengram_tools(api_key="om-...")
# → 5 tools: search, remember, profile, save_workflow, workflow_feedback

agent = Agent(role="Support Engineer", tools=tools)

Agent Templates

Ready-to-run examples — clone, set API key, run in 5 minutes:

Template	Stack	What it shows
DevOps Agent	Python SDK	Procedures that evolve from deployment failures
Customer Support	CrewAI	Agent with 5 memory tools, remembers returning customers
Personal Assistant	LangChain	Cognitive profile + auto-saving chat history

cd examples/devops-agent && pip install -r requirements.txt
export MENGRAM_API_KEY=om-...
python main.py

API Reference

All endpoints require Authorization: Bearer om-... — your key identifies you, no user_id needed.

Endpoint	Description
POST /v1/add	Add memories (auto-extracts all 3 types)
POST /v1/search	Semantic search
POST /v1/search/all	Unified search (all 3 types)
GET /v1/episodes/search	Search episodic memories
GET /v1/procedures/search	Search procedural memories
PATCH /v1/procedures/{id}/feedback	Report success/failure → triggers evolution
GET /v1/procedures/{id}/history	Version history + evolution log
GET /v1/profile	Cognitive Profile
GET /v1/triggers	Smart Triggers (reminders, contradictions, patterns)
POST /v1/agents/run	Run memory agents (Curator, Connector, Digest)

Full interactive docs: mengram.io/docs

License

Apache 2.0 — free for commercial use.

---

Built by Ali Baizhanov · mengram.io