Optional Memory Booster

Make Your Frenzy.bot Remember More

Graphiti is a simple add‑on that gives your Frenzy.bot long‑term memory. It helps your assistant remember past chats, documents, and decisions so replies feel smarter and more personal.

What Is Graphiti for Frenzy.bot?

Graphiti is a memory layer we can add to your Frenzy.bot. It captures key moments from chats, docs, and tickets, then stores them so the bot can use them later.

That means better answers over time—without you changing prompts or repeating yourself. It’s an optional upgrade, not required.

Why This Makes Your Bot Better

Your bot remembers more, so customers repeat less.

Graphiti adds long‑term memory so your Frenzy.bot can keep context across time.

Remembers customer preferences and past tickets

Connects earlier chats to today’s questions

Finds the right info faster, even in large files

Gives more consistent, personalized answers

Key Capabilities

Memory Ingestion (Episodes)

Graphiti ingests conversations, summaries, and structured JSON as "episodes." Each episode becomes structured knowledge in the graph.

User onboarding history Support ticket context Sales discovery data Meeting summaries Product requirements
{
  "episode_id": "ep_001",
  "source": "customer_chat",
  "content": "User prefers AWS over GCP...",
  "timestamp": "2026-01-22T16:56:00Z"
}

“What are this user’s preferences?”

Graphiti returns the exact preference nodes and supporting facts.

Semantic Search Across Facts

Graphiti doesn’t just search text. It searches facts and relationships. Ask a question and Graphiti finds the most relevant nodes, facts, and edges.

Temporal Knowledge Graph

Graphiti models change over time. You don’t just get static facts—you get how facts evolved.

“How has the customer’s requirement changed in the last month?”

Entity + Relationship Extraction

Graphiti automatically extracts people, companies, products, requirements, decisions, events, locations, and more. No manual tagging required.

Always-On, Low Latency

Graphiti is designed for real-time memory retrieval. It integrates directly into the chat pipeline so answers are generated with memory in the loop.

Where the Add‑On Helps Most

Customer Support

Remember previous tickets, resolutions, and user preferences. Reduce repetition and shorten response time.

Sales & Onboarding

Track business requirements across multiple conversations. Keep a running “mental model” of every account.

Internal Knowledge

Build a living knowledge layer from Slack, docs, and meetings. Query institutional memory instantly.

Product Teams

Capture feedback and requirements as structured facts. See what changed, when, and why.

How It Works

1

Capture

Conversations and events are ingested as episodes.

2

Extract

Entities, relationships, and facts are automatically identified.

3

Store

Knowledge is stored in a graph database (FalkorDB or Neo4j).

4

Retrieve

Relevant facts are injected into AI responses in milliseconds.

What’s Under the Hood (Optional)

  • MCP-compatible server (HTTP transport)
  • FalkorDB default; Neo4j supported
  • LLM-agnostic (OpenAI, OpenRouter, Anthropic, Gemini, Groq)
  • Tooling: add_episode, search_nodes, search_facts, get_status, etc.
  • Telemetry opt-out supported

Why Teams Choose This Add‑On

Persistent memory without retraining
Structured knowledge, not just embeddings
Works with your existing AI stack
Fast, lightweight, production-ready

Security & Privacy

Graphiti never stores raw API keys or personal data in telemetry.

You control deployment, storage, and access policies.

"Your Frenzy.bot stops feeling forgetful and starts acting like a real assistant that remembers what matters."