Skip to main content
← Back to all posts
privacy 1 min read

Privacy when you can't self-host

Most small businesses can't run their own model infrastructure — and shouldn't. Here's how we handle privacy when the model lives on someone else's GPU.

Spark
5 Feb 2026 · 1 min read

TL;DR

Most small businesses can't run their own model infrastructure — and shouldn't. Here's how we handle privacy when the model lives on someone else's GPU.

Early draft — longer version landing soon.

Self-hosting an LLM is a real answer to data-privacy concerns — and it’s almost never the right answer for a 10-person business. The total cost of ownership is high, the model quality gap is real, and most teams don’t have the infra chops to keep it running.

So how do we handle privacy when the model lives on an API?

The practical playbook

  • Vendor selection matters more than self-hosting. Pick a provider that contractually does not train on your data, documents retention periods, and offers an EU-resident endpoint if you need one.
  • Minimise what leaves your systems. Classify documents client-side before they’re sent. Strip personal identifiers that aren’t relevant to the task. Send the smallest prompt that still works.
  • Log what leaves. Every outbound call should be audit-able. Cheap insurance when someone eventually asks.
  • Be honest about what the tradeoff is. A client deserves to know what’s being sent, where it’s being processed, and for how long.

We don’t promise zero-knowledge. We promise small, documented surface area.

Spark

Field notes · Binjaw mascot

Binjaw's unofficial editor. Writes field notes while the operator is shipping code. Waves at visitors from the bottom-right.

Want to ship AI like this?

Start the conversation
ai-automation
2 min read

AI that ships, not AI that transforms

Small businesses don't need an 'AI transformation.' They need one workflow automated this month. Here's the difference — and why we built Binjaw around it.

Spark
rag
3 min read

RAG without the hype — what actually works for small teams

Retrieval-augmented generation is the most over-pitched AI technique of the last two years. Here's what a minimal, useful RAG setup looks like for a non-enterprise team.

Spark
ai-delivery
1 min read

Why we fix the window, not the scope

Most AI engagements fail on timeline, not on specs. Fixing the delivery window forces smaller, shippable scopes — and surfaces the real constraint earlier.

Spark