Bring Your Own LLM: Use Your Mac's AI Capabilities With Launch My AI

If you already run Launch My AI, you know the deal: you own your own private assistant on a server you control, Dex keeps that server healthy, and you can choose to never touch a terminal. Until now, your assistant has done its thinking using Anthropic's Claude, which you pay for directly based on your usage.

With Bring Your Own LLM, you now have another option. If you own an Apple Silicon Mac (an M-series MacBook, Mac mini, Mac Studio, or iMac), you can point your assistant at an AI model running on your own Mac at home instead of Claude.

This is the first article in a 9-part series that walks you through the whole thing, step by step, in plain English. This page is the map. Later articles go deep on each part. Start here to understand what BYOLLM is, whether it's right for you, and what to expect before you commit.

What "Bring Your Own LLM" (BYOLLM) actually means

Your Launch My AI assistant needs an "engine" to generate its answers. Think of that engine as the brain doing the reasoning behind every reply.

By default, that brain lives in the cloud (Claude). With Bring Your Own LLM, the brain instead lives on a computer you already own and keep at home: your Apple Silicon Mac. Your assistant sends its questions to your Mac, your Mac does the thinking, and the answer comes back.

Nothing else about your setup changes. You still get:

• Your own private assistant on your own server
• Dex quietly keeping that server running
• No Linux, no terminal, no technical babysitting on the server side

It simply swaps where the thinking happens.

Why would you want this?

This feature isn't for everyone, and that's fine. But for the right person, it's genuinely appealing.

1. Keeping AI conversations on hardware you own

For a lot of our customers, the whole reason they came to Launch My AI was privacy and ownership. This takes that a step further: when your Mac is doing the thinking, your assistant's reasoning happens on a machine sitting in your own home or office, not a third-party AI provider.

We'll always be honest about what this does and doesn't mean. If you choose to install your assistant on a Digital Ocean Droplet, your assistant still lives on your cloud server, and data still travels between that server and your Mac. But the model itself — the "brain" — runs on your hardware, and that matters to a lot of people.

If you choose the 'Bring Your Own Device' option your assistant also lives on a device running on your home network, and in that situation your data will never leave your home network! You can even chose to have your assistant running on the same Mac as your LLM!

2. Avoiding per-use AI charges

With Claude, you pay Anthropic based on how much you use. Some people love the simplicity; others would rather run a model they already have the hardware for and skip usage-based charges for that model. If you already own a capable Mac, this lets you put it to work to avoid paying token fees to Anthropic for your assistant.

3. Experimenting and learning

Some folks just want to try different AI models and see how they compare. This opens that door without turning you into a system administrator.

Who this is (and isn't) for

Bring Your Own LLM is a good fit if you:

• Own an Apple Silicon Mac (M1 or newer) with a reasonable amount of memory
• Are comfortable keeping that Mac on and connected when you want your assistant to use it
• Care about running the AI model on hardware you control

**It's probably not the right choice if you:**

• Don't own an Apple Silicon Mac (Intel Macs and other computers aren't supported by this feature)
• Want your assistant available even when your home computer is asleep or off
• Prefer the "set it and forget it" simplicity of the default cloud option

That last point is worth sitting with. When your assistant's brain lives on your Mac, your assistant depends on that Mac being awake and reachable. Close the lid, lose your internet, or shut it down, and the brain goes quiet until it's back. This is a real trade-off, not a footnote — and we'll cover how to manage it later in the series.

Honest expectations about running the AI on your Mac

We want you to go in clear-eyed, because a good experience starts with realistic expectations.

• A model on your Mac behaves differently than Claude. It may answer in a different style, handle some tasks better or worse, and have different limits. This is normal. Part of the series is about helping you pick a model that fits your needs and your Mac's capabilities.
• Your Mac's specs matter. More memory generally means you can run larger, more capable models. A base MacBook Air and a maxed-out Mac Studio are very different experiences.
• It's provided as-is. As with everything at Launch My AI, we don't promise specific accuracy, speed, or results — and when the model runs on your Mac, a lot of the performance depends on your own hardware and setup. We'll give you the best guidance we can, honestly.

None of this is meant to scare you off. Plenty of people will love this option. We just believe the right customers are informed ones.

You can switch back anytime

Here's the reassuring part: this isn't a one-way door. If you set up your Mac as your assistant's brain and later decide it's more hassle than it's worth — or you're heading out of town and need to take your Mac with you and want your assistant to keep working — you can point it back to the default cloud option. The switch is designed to leave your assistant, your server, and your data in place.

Think of it as a setting you can flip based on what you need that week, not a permanent commitment.

What the rest of this series covers

This overview is Part 1. The remaining eight articles get hands-on:

• Checking whether your Mac is up to the task
• Choosing a model that fits your hardware and goals
• Getting the model running on your Mac
• Connecting your Mac to your Launch My AI assistant
• Keeping the connection reliable while your Mac is in use
• Understanding the trade-offs around availability and performance
• Troubleshooting common hiccups
• Deciding when to use your Mac versus the cloud option

You don't need any technical background to follow along. If you can install an app and follow clear instructions, you can do this.

A quick reminder about the essentials

If you're brand new to Launch My AI: our portal, the Launchpad, lets non-technical people launch their own private assistant on a cloud server they own, with Dex handling the upkeep so you never touch a terminal.

The Starter plan is normally $9.99/month, and right now our launch offer is $5.99/month for your first 6 months (40% off), then it renews at $9.99/month. If you choose to install your assistant in the cloud, you also pay your own cloud provider (roughly $12/month) directly, and — if you use the default cloud brain — you pay for your own Anthropic usage. Bring Your Own LLM is simply another way to power your assistant once it's up and running.

Ready to start?

If you own an Apple Silicon Mac and the idea of running your assistant's brain on your own hardware appeals to you, you're in the right place. Head to Part 2, where we'll help you figure out exactly what your Mac can handle before you download anything.

Own your assistant. Own your server. And now, if you like — own the brain, too.