Build AI assistants that truly know their users. Remember preferences, habits, and context across every conversation.
Personal assistants forget everything between sessions. Users re-explain preferences, projects, and context every time.
Without memory, the assistant gives generic responses that don't reflect the user's unique needs and style.
The assistant can't recognize patterns in the user's behavior — daily routines, recurring tasks, or preferred workflows.
AI companions feel shallow because they don't accumulate shared experiences or inside references.
Semantic memory stores preferences, interests, and personal details. The AI knows the user inside and out.
Episodic memory remembers conversations, decisions, and events. The AI references past interactions naturally.
Procedural memory captures daily workflows, recurring tasks, and preferred processes that evolve over time.
One API call generates a system prompt with the user's full context — making every LLM instantly personalized.
from mengram import Mengram
m = Mengram(api_key="mg-...")
# Morning check-in — AI remembers everything
profile = m.profile(user_id="alice")
# "Alice is a product manager at Acme Corp. She prefers morning standup
# summaries with bullet points. She's working on the Q1 launch...
# Yesterday she reviewed the design specs and had feedback on the nav..."
# After each conversation, memory grows
m.add("Alice asked me to remind her about the design review on Friday. "
"She also mentioned she prefers Figma links over screenshots.",
user_id="alice")
# Next session: the AI remembers the reminder and preference
memories = m.search("design review", user_id="alice")Context retention
Session continuity
Memory depth
Free API key. No credit card required. Start in 60 seconds.