Memori

Use Cases

Memori is designed for any application where AI agents need to remember context across conversations and agent executions. Here are the most common use cases.

Customer Support Chatbots

Build support bots that remember customer history, preferences, and past issues. No more "Can you repeat your account number?" — Memori recalls context automatically.

Benefits:

  • Remember customer preferences and history
  • Recall previous support tickets and resolutions
  • Personalize responses based on past interactions
  • Track issues across multiple sessions
Customer Support
from memori import Memori
from openai import OpenAI

client = OpenAI()
mem = Memori().llm.register(client)

# Each customer gets their own memory space
mem.attribution(
    entity_id="customer_456",
    process_id="support_bot"
)

# Memori automatically recalls relevant context
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{
        "role": "user",
        "content": "I'm having that issue again"
    }]
)
# Memori injects: "Customer previously reported
# login timeout issues on 2024-01-15"

Personalized AI Assistants

Create AI assistants that learn and adapt to each user over time. Memori builds a profile of preferences, skills, and context that makes every interaction more relevant.

Benefits:

  • Learn coding preferences and tech stack
  • Remember project context across sessions
  • Adapt communication style to user preferences
  • Build long-term user profiles automatically
Personalized AI
from memori import Memori
from anthropic import Anthropic

client = Anthropic()
mem = Memori().llm.register(client)

mem.attribution(
    entity_id="developer_789",
    process_id="code_assistant"
)

# Over time, Memori learns:
# - "Uses Python 3.12 with FastAPI"
# - "Prefers type hints and dataclasses"
# - "Works on e-commerce platform"
response = client.messages.create(
    model="claude-sonnet-4-5-20250929",
    max_tokens=1024,
    messages=[{
        "role": "user",
        "content": "How should I structure this endpoint?"
    }]
)

Multi-Agent Workflows

Coordinate multiple AI agents that share context through Memori. Each agent contributes to a shared memory space while maintaining its own process identity and conversation history.

Benefits:

  • Share context between specialized agents
  • Track which agent contributed what information
  • Maintain conversation and execution continuity across handoffs
  • Build collective knowledge graphs
Multi-Agent
from memori import Memori
from openai import OpenAI

client = OpenAI()
mem = Memori().llm.register(client)

# Research agent gathers information
mem.attribution(
    entity_id="project_alpha",
    process_id="research_agent"
)
client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{
        "role": "user",
        "content": "Research competitor pricing"
    }]
)

# Analysis agent recalls research findings
mem.attribution(
    entity_id="project_alpha",
    process_id="analysis_agent"
)
# Memori shares context across agents
# for the same entity

Enterprise IT Operations — Incident Response

Large enterprises run hundreds of services across complex infrastructure. When incidents strike, response time is critical. Memori captures every tool call, diagnostic decision, and resolution outcome as structured trace memory — so agents accumulate institutional knowledge across incidents and come back faster each time.

Benefits:

  • Recall past incidents with similar error patterns and known resolutions
  • Build a persistent knowledge base of system behavior across thousands of incidents
  • Reduce mean time to resolution by surfacing what worked before
  • Audit the full decision trail: which tools ran, what they returned, and what action was taken
  • Route incident context across specialized agents — triage, escalation, and remediation
Enterprise IT Operations
import os
from memori import Memori
from openai import OpenAI

client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
mem = Memori().llm.register(client)

# Each service gets its own memory space; the agent is the process
mem.attribution(
    entity_id="payment-service-prod",
    process_id="incident_response_agent"
)

tools = [
    {
        "type": "function",
        "function": {
            "name": "query_logs",
            "description": "Query application logs for a time range and filter",
            "parameters": {
                "type": "object",
                "properties": {
                    "service": {"type": "string"},
                    "time_range": {"type": "string"},
                    "filter": {"type": "string"}
                },
                "required": ["service", "time_range"]
            }
        }
    },
    {
        "type": "function",
        "function": {
            "name": "get_metrics",
            "description": "Retrieve service metrics from the monitoring system",
            "parameters": {
                "type": "object",
                "properties": {
                    "service": {"type": "string"},
                    "metric": {"type": "string"}
                },
                "required": ["service", "metric"]
            }
        }
    },
    {
        "type": "function",
        "function": {
            "name": "restart_pod",
            "description": "Restart a service pod in the specified region",
            "parameters": {
                "type": "object",
                "properties": {
                    "service": {"type": "string"},
                    "region": {"type": "string"}
                },
                "required": ["service", "region"]
            }
        }
    }
]

# Memori intercepts this call — tool calls, results, and decisions
# are captured as trace events and converted into structured memory
response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[
        {
            "role": "system",
            "content": "You are an enterprise IT operations agent. Diagnose and resolve infrastructure incidents."
        },
        {
            "role": "user",
            "content": (
                "P1 Alert: payment-service-prod returning 503 errors. "
                "Error rate 42%, p99 latency 8.2s. Started 14 minutes ago. "
                "Diagnose and resolve."
            )
        }
    ],
    tools=tools
)

# After resolution, Memori has stored:
# - The alert conditions that triggered the incident
# - Every tool call made and what it returned
# - The diagnostic path and decisions taken
# - The resolution action and outcome
#
# Next incident: the agent automatically recalls
# "Last time payment-service-prod had 503s with elevated p99 latency,
#  logs showed DB connection pool exhaustion — resolved by restarting
#  the us-east-1 pod. Resolution time: 6 minutes."