We're Betting on Local AI. The Industry Is Catching Up.

We've been building something at Toodlebox that felt contrarian: instead of asking customers to trust any cloud with their data, we've been working toward AI that runs on hardware they control. The reasoning is simple - governance has to live where the data lives. If your AI's decision logs and business data sit on hardware inside your building, the boundary is the building.

It's been a lonely argument. But it's starting to feel inevitable.

Meta just released Muse Glimmer, an open-weight 30B parameter model designed to run locally on consumer GPUs. Zuckerberg published a 6,000-word essay arguing the future of AI should be decentralized, not locked in a handful of proprietary providers. Perplexity launched Hybrid Compute - keeping sensitive work on-device, sending reasoning to the cloud. Nvidia is buying Hugging Face for $13 billion, betting the entire open-weight ecosystem isn't going anywhere.

We didn't build this way because we predicted the market would move here. We built this way because the businesses we serve - small businesses who'd otherwise lean on an outside provider just to keep their IT running - are going to need their AI on hardware they control. That was a design constraint driven by who we're building for.

But the convergence is getting harder to ignore. The biggest players are all drifting toward some version of "run it locally, use the cloud when it's convenient." They're starting from the data center and working backward. We're starting from the customer's office and working forward.

When Meta or Perplexity say "local," they mean your laptop running their app. That's a good start. But we think local means something more - and we'll have more to say about that soon.

Every open-weight model that drops - Glimmer today, whatever comes next - makes the case stronger. A real model running on a consumer GPU is going to change the math on what a small business can do without trusting a stranger's cloud. It hasn't happened yet. But the pieces are all moving into place.