Building in public: EdgeAI — your AI that knows your contacts

· Meaningful Blog

We built an AI that actually stays private

Most AI features today are wrappers around OpenAI or Anthropic APIs.

Your data leaves your app, hits a third-party server, and comes back wrapped in a response. You hope they don't train on it.

We wanted something different.

---

Meet EdgeAI

EdgeAI is Meaningful's built-in AI assistant. It lives entirely within our infrastructure — hosted on a dedicated cloud server that we control.

No OpenAI. No Anthropic. No public API calls. Your conversations, your connections, your journal entries — none of it leaves our ecosystem.

What it can do

  • Chat naturally — ask about your connections, upcoming events, or journal entries in plain language
  • Add connections, events, and journal entries — just tell it what you need, it handles the rest
  • Surface insights — stale connections, pending commitments, network health
  • Remember you — it learns your preferences, goals, and relationships over time
  • Answer general questions — it's a real LLM, not a keyword matcher

What makes it different

  • Private by design — runs on our own Ollama server, not a public API
  • Encrypted at rest — personal data is AES-256 encrypted in the database
  • No training on your data — the model is open-source (Llama 3.2), we don't fine-tune on user data
  • Works offline-capable — in local mode, it runs entirely on your machine

---

Why this matters

AI is only useful if you trust it with real information.

If you can't tell your assistant about your actual relationships, goals, and commitments — it's just a chatbot.

EdgeAI is built so you can.

Have you tried an AI tool that actually respects your privacy? What was the experience like?

---

See EdgeAI in Action

Here's a quick demo of EdgeAI handling natural commands and managing your relationships:

Your browser does not support the video tag.