Hi HN, I'm Ali. I've been building Mengram for the past year.

  The problem: Every AI memory tool stores facts — "user likes dark mode." But when my agents failed at a task, they'd fail the exact same way next time. They had no memory of what happened or how to do things better.                                                               
                                                                                                                                                                                                                                                                                          
  What Mengram does: It stores 3 types of memory, modeled after how human cognition works:                                                                                                                                                                                                
                                                                                                                                                                                                                                                                                          
  - Semantic — facts and preferences (like Mem0, Zep)                                                                                                                                                                                                                                     
  - Episodic — events, decisions, outcomes (what happened and when)                                                                                                                                                                                                                       
  - Procedural — learned workflows that evolve when they fail                                                                                                                                                                                                                             
                                                                                                                                                                                                                                                                                          
  The procedural part is what I'm most excited about. When an agent reports a failure, Mengram automatically evolves the procedure — adds a new step, changes the order, removes what didn't work. Your agent literally gets better at its job over time.
                                                                                                                                                                                                                                                                                          
  Example: Week 1, "Deploy" = build → push → deploy. Agent forgets migrations, DB crashes. Week 2, Mengram evolves it to build → run migrations → push → deploy. Agent hits OOM. Week 3, adds memory check step. This happens automatically.                                              
                                                                                                                                                                                                                                                                                          
  Technical details:                                                                                                                                                                                                                                                                      
  - Python & JS SDKs: pip install mengram-ai                                                                                                                                                                                                                                              
  - Free cloud API (no credit card) or fully self-hostable
  - MCP server for Claude Desktop / Cursor (21 tools)
  - LangChain, CrewAI, OpenClaw integrations
  - Knowledge graph + vector search + reranking
  - Cognitive Profile: one API call generates a system prompt from all memories

  What it's NOT good at (yet):
  - Smaller community than Mem0 (they have 25K stars, I'm just starting)
  - No SOC2/HIPAA yet (Zep has this)
  - No agent-controlled memory like Letta/MemGPT

  I'd love feedback on the API design and the procedural memory concept. Is this something you'd actually use in production?

  GitHub: https://github.com/alibaizhanov/mengram
  Docs: https://mengram.io/docs
  Get a free key: https://mengram.io