surfacevector Get Horus

Local AI sovereignty

Keep the model on your side of the wire.

Every prompt you send to a hosted model crosses a boundary you don't own. Running the model yourself moves that boundary back to your hardware. Here's what that changes, what it costs, and what it doesn't fix.

Definition

Sovereignty is four separate controls

"Private AI" gets used loosely. It's worth separating what you actually hold onto, because a system can give you one of these and none of the others.

Data
What the model reads, and where it stops. Prompts, attached files, source code, customer records, log excerpts. On a local system all of it stays on the machine doing the work — there is no copy on someone else's disk to retain, subpoena, or leak.
Access
What the system is allowed to touch. Which directories, which hosts, which credentials, which network segments. You set the reach, and you can prove what it was by reading the config yourself.
Behavior
Which model, which version, which rules. A hosted model can change under you overnight and quietly break a workflow you tuned around it. Local weights don't move unless you move them.
Continuity
Whether it still works next quarter. No deprecation notice, no rate limit, no price change, no outage on someone else's status page. The system you set up is the system you keep.
Comparison

Where your data goes, side by side

Neither column is wrong for every job. The point is to know which one you're choosing.

Hosted API vs. local model — the practical differences
 Hosted APILocal model
Prompt contentsLeave your network on every requestNever leave the machine
Files and contextUploaded to a third partyRead from your own disk
RetentionGoverned by a policy you can read but not enforceWhatever your own logging keeps
AvailabilityDepends on their uptime and your linkWorks with the network unplugged
Cost shapeMetered per token, foreverHardware upfront, then electricity
Model changesCan shift without warningOnly when you pull new weights
Rate limitsEnforced by the providerBounded by your own GPU
Peak capabilityFrontier-class on the hardest tasksStrong, and improving fast, but not frontier
Who maintains itThey doYou do
Honest limits

What local AI doesn't fix

Anyone selling you local AI without this section is selling. Four things move onto your plate the moment you bring the model home.

01

The box is now the asset

You removed the exposure of data in transit and replaced it with a machine holding everything, sitting on your LAN. Disk encryption, account separation, patching, and physical access all matter more than they did before.

02

You're the one on call

Driver updates, VRAM exhaustion, a model that stops calling tools correctly after an upgrade — nobody else is going to notice or roll it back. Budget the maintenance honestly before you commit a workflow to it.

03

There is still a capability ceiling

Open-weight models have closed most of the everyday gap and keep closing it. On the hardest long-horizon reasoning they are not yet frontier. Match the model to the task instead of forcing one tool onto every job.

04

Local doesn't mean careful

A model with shell access on your own hardware can still delete the wrong directory. Sovereignty is about who holds the controls, not about whether the controls are needed — approval gates, scoped permissions, and backups still do the real work.

Use cases

Who runs local, and why

Sensitive work

Data that can't leave

Client records, case files, health or financial detail, internal source code. When the rule is "this doesn't go to a third party," a local model is the only version of the tool you're allowed to use.

Small business

Predictable cost

One machine, one power bill, no per-seat pricing and no surprise invoice when usage doubles. The tool you bought stays the tool you own.

Builders

Full access to the stack

Swap models, tune sampling, wire in your own tools, run the whole loop offline, and read every log. Nothing is behind an endpoint you can't inspect.

Disconnected

Works without a link

Remote sites, restricted networks, boats, vehicles, and anywhere the connection is slow, metered, or not permitted. Inference doesn't care that the uplink is down.

Specifications

What it actually takes to run

The GPU is the constraint that matters. Everything else is about keeping it fed. These are general guidelines for local inference, not a hard floor.

VRAM class, and what it gets you
VRAMModel class it fitsRealistic use
8 GB7–8B at 4-bit quantizationDrafting, summarizing, quick code help, short context
12–16 GB12–14B comfortably, or 8B with long contextDaily assistant work with room for tools and documents
24 GB24–32B at 4-bitThe common sweet spot for agentic, tool-calling workloads
48 GB+70B class, or several models residentHeavier reasoning, parallel jobs, vision plus text together
System RAM
32 GB is a comfortable floor; 64 GB gives room to run other work alongside the model. If layers spill out of VRAM, system memory absorbs them — slowly.
Processor
A modern multi-core CPU. It won't set your token rate when the model fits on the GPU, but it handles prompt processing, tooling, and everything the agent runs outside the model.
Storage
NVMe, with real headroom. A single quantized model file runs from a few gigabytes to forty, and you will end up keeping more than one. Fast storage also cuts load time between model swaps.
Operating system
Any modern 64-bit OS. Linux is the path of least resistance for drivers, containers, and unattended service management.
Vocabulary

The terms you'll hit in the first hour

Inference
Running a trained model to get an answer. This is what your GPU is doing; training is a separate, far heavier job you're not signing up for.
Weights
The model file itself. Open-weight models are ones you can download and run without asking permission per request.
Quantization
Compressing weights to lower precision so they fit in less VRAM. Labels like Q4_K_M describe how aggressively. Lower precision means smaller and faster, with some quality traded away.
VRAM
Memory on the graphics card. The single biggest factor in which models you can run and how fast they respond.
Offloading
Pushing layers that don't fit in VRAM into system RAM. It keeps a big model running, and it is dramatically slower.
Context window
How much text the model can hold at once — your prompt, the conversation, and any files it read. Bigger windows cost memory.
Token
The chunk a model reads and writes in, roughly a short word or word-fragment. Speed is measured in tokens per second.
Tool calling
The model emitting a structured request to run something — read a file, run a command, query an API. It's the difference between a chatbot and an operator.
RAG
Retrieval-augmented generation: searching your own documents and handing the relevant passages to the model, so it answers from your material instead of memory.
Egress
Data leaving your network. The metric this whole page is about.
Air gap
A system with no network path outward at all. Local inference is what makes an air-gapped AI workflow possible.
Agentic
A model working in a loop — plan, act with tools, read the result, adjust — instead of answering once and stopping.
Questions

Frequently asked questions

What is AI sovereignty?
AI sovereignty means the model, the data it reads, and the decisions about how it behaves all sit inside a boundary you control. In practice: inference runs on hardware you own, you choose the model and its version, and you decide what the system is allowed to touch.
Why run AI locally instead of using a hosted API?
A hosted API is a boundary crossing. Every prompt, file, and code snippet you send leaves your network and lands on infrastructure governed by someone else's retention policy, pricing, and update schedule. Running the model locally removes that hop entirely.
Is a local model as capable as a hosted frontier model?
For most everyday work — drafting, summarizing, code assistance, structured extraction, tool-driven automation — a well-chosen open-weight model in the 24B to 32B class is close enough that the gap rarely shows. For the hardest long-horizon reasoning, hosted frontier models still lead. Choose per task, not per ideology.
What hardware do I need to run a local AI system like Horus?
The GPU is the constraint that matters. Roughly 8 GB of VRAM runs small models at 4-bit quantization; 24 GB comfortably runs the 24B to 32B class that agentic and tool-calling work tends to need; 48 GB or more keeps several models resident at once. Pair it with 32 GB or more of system RAM, a modern multi-core CPU, and NVMe storage with room for several model files.
Does local AI work without an internet connection?
Yes. Once the weights are on disk, inference needs no network at all. Anything that reaches outward — web search, email, remote hosts — still needs a connection, but the model itself keeps working when the link drops.
Is local AI automatically more secure?
No. It removes one exposure — your data leaving the network — and hands you the rest of the job. The machine running the model becomes the asset to protect: disk encryption, access control, patching, logging, and backups are yours to run. Local is a different risk profile, not a smaller one.
What does running local AI cost?
The cost moves from a metered per-token bill to hardware and electricity. There is a real upfront number for a capable GPU and no per-request charge after it, which favors heavy or sustained use. Light, occasional use is usually cheaper on a hosted API.
How does Horus fit into this?
Horus is a local-first AI operator from Surface Vector Technologies. It runs on your hardware, keeps its work inside your network by design, and goes past chat: it drives files, shells, browsers, and scheduled jobs, with human approval before it changes anything.

Surface Vector Technologies

Horus: an AI operator that stays on your hardware

Most local AI setups stop at chat. Horus is built to do the work — read and edit files, run commands, drive a browser, and hold a job through to the end — while the whole loop stays inside your network.

  • Runs on your machine, with no cloud egress on the inference path
  • Agentic tool use across files, shell, browser, and scheduled jobs
  • Human approval before it changes anything on disk
  • Bring your own open-weight models and swap them whenever you want
See what Horus does →