Capability	What it does
🧠 Memory	Save facts, search them semantically across sessions
🌐 Web savings	Compressed context — fewer tokens to the LLM
💬 Thread	Long conversations compressed (-72%), details restored on demand
📊 Usage	See your real token savings (today + all-time)

Install (community node)

# In your n8n folder:
npm install n8n-nodes-vibo
# Restart n8n → the "ViBo Memory" node appears in the canvas

Or install from the n8n Editor: Settings → Community Nodes → Install → n8n-nodes-vibo

Setup

Get a key: https://wwwvibo.com — free tier: 500 facts forever, then $5/mo · $30/yr · $60 lifetime (https://wwwvibo.com/pricing).
In n8n: create ViBo API credential, paste your key (starts with VIBO-).
Add the ViBo Memory node to your workflow.

Operations

Memory → Search

Find relevant facts in your memory. Returns facts + savings stats:

{
  "ok": true,
  "facts": [{ "label": "client-anna", "content": "Anna likes coffee, order #42", "level": "L1" }],
  "saved_tokens": 13452,
  "saved_pct": 97.5
}

Memory → Add

Save a fact (deduplicated by exact match):

{ "ok": true, "added": true, "nodes": 45 }

Memory → Usage

Your real savings (today + all-time):

{
  "ok": true,
  "records": 21,
  "saved_tokens": 8428443,
  "saved_usd": 1.18
}

Thread → Add / Compress / Ask / Context

Conversation memory: add messages, compress old ones (-72%), ask "what did we discuss 3 days ago?", get compressed context for the LLM.

Usage examples

Support bot that remembers clients

[Telegram trigger] → [ViBo Memory: Search "client X"] → [AI Agent] → [Reply]
                        ↓ (no result? add fact)
                    [ViBo Memory: Add "client X likes email"]

Conversation memory for your AI agent

[Chat trigger] → [ViBo Memory: Thread Add (user msg)] → [AI Agent]
                     ↑                          ↓
                     [ViBo Memory: Thread Context] ← chat history

Show your client the savings

[Any trigger] → [ViBo Memory: Usage] → [Message: "You saved X tokens"]

API

The node talks to https://wwwvibo.com:

Endpoint	Operation
`POST /memory/search`	Find relevant facts (returns savings stats)
`POST /memory/add`	Save a fact (dedup by exact match)
`POST /memory/thread`	Thread memory: add / compress / ask / context
`GET /usage?key=`	Token savings statistics

Security: data encrypted (L1/L2/L3), per-key memory, key verification on every call. Secrets never reach the LLM.

Honest about savings

Memory: 97.5% fewer tokens on 118 facts (grows with memory, up to 50-150× on 10K+).
Web: 96-99% (measured 47,443 → 186 tokens).
Threads: -72%.
Code generation: little to no savings (it doesn't "remember").
Every operation reports its real savings — no marketing numbers.