For the last weeks, I’ve been building a local-first AI desktop companion that lives on your screen.

It can:

• see your screen
• understand your context
• execute actions on your system

But it had one big limitation:

It only lived on your desktop

So I changed that.

🌐 Introducing: Blob Connectors

I just added a new layer to OpenBlob:

👉 Blob Connectors

A lightweight Python bridge that connects your local AI to the outside world:

• Telegram
• Discord
• Slack
• Email

🧠 What this actually means

You can now do things like:

• send open spotify via Telegram → Spotify opens on your PC
• ask a question in Discord → your local model answers
• send an email → get a contextual AI reply
• control your desktop from anywhere

And the important part:

It’s still local-first

⚙️ How it works

All channels go through the same pipeline:

Telegram / Discord / Slack / Email
              │
        Blob Connectors (Python)
              │
    ┌─────────┴─────────┐
    │                   │
OpenBlob running?    Ollama fallback
(localhost)         (local model)
              │
        Command Router
              │
      Desktop action

Everything becomes a normalized Message object.

No matter where it comes from.

🔌 Why this matters

This is not just “adding integrations”.

This is the first real step towards:

an AI system that exists beyond a single interface

Now OpenBlob is:

• not just UI-bound
• not just voice-bound
• not just desktop-bound

It becomes a distributed interface to your own system

🧩 Built for extension

Each connector implements the same interface:

class MyConnector(BlobConnector):
    async def receive_message(self, raw) -> Message | None: ...
    async def send_response(self, original: Message, response: str) -> None: ...
    async def start(self) -> None: ...

So adding new platforms is trivial:

• WhatsApp
• Matrix
• iMessage (maybe 👀)
• anything with an API

🔒 Still local-first

Important:

• runs on your machine
• uses your local models (Ollama)
• no required cloud backend
• transparent behavior

If OpenBlob is offline:

→ it automatically falls back to local reasoning

🚧 Current state

• works across all channels
• still early
• structure is stabilizing
• lots of room for improvement

🔮 What this unlocks next

This connector layer enables things like:

• shared memory across all channels
• persistent conversations
• multi-agent systems
• calendar / tool integrations
• real remote control of your system

🤝 If you want to build with me

This is probably the best moment to jump in.

You can:

• build new connectors
• improve routing / memory
• design better UX
• experiment with AI behaviors

👉 https://github.com/southy404/openblob

💡 Final thoughts

This is mainly an infrastructure update.

By introducing a connector layer and a normalized message interface, OpenBlob becomes:

• easier to extend
• easier to integrate
• less tied to a single UI

It’s a small surface change — but a significant internal shift.