Local-first · Private by design

Your computer. Your AI.
No data center required.

Run AI on the hardware you already own — fully local, or turn your home machine into a personal AI server you reach from anywhere. Every model, every device, yours.

Run 100% localWe never store your requestsAny model, your hardware

Beta opens July 4 —

Central-Intel
The Central-Intel desktop console — a local-first AI dashboard showing requests, token usage, savings, and on-device model activity.

An AI that's actually yours

The big assistants run on their hardware, their model, their account. Central-Intel flips all three.

Be your own server

Run AI fully local, or turn your home desktop into a personal AI server you reach from your laptop or phone anywhere. The thinking happens on a machine you own — no hyperscaler data center.

Any model, one endpoint

Claude, GPT, Gemini, or a local model — swap freely behind a single OpenAI-compatible address. You're never welded to whichever giant made your phone.

A memory that's yours

Persistent memory, profiles, and your personal knowledge base enrich every request — stored encrypted on your device, not in someone else's cloud.

Local AI in clicks

No terminal, no compatibility headaches. Install and go — Central-Intel sets up local AI for you and grows with you as you're ready.

It remembers you

Open a brand-new chat and it already knows what matters — pulled from past conversations, stored on your device. No re-explaining, ever.

Central-Intel

Your private AI command center

Memory, knowledge, profiles, and routing — one local-first console that's loyal to you alone.

Memory

It actually remembers you

Every fact worth keeping is pulled from your conversations and quietly reused — so you never re-explain yourself. Stored encrypted, on your device.

Central-Intel Memories panel — facts automatically learned about you, organized by category.
Profiles

A different mind for every job

Switch personality, model, and knowledge with a click.

Central-Intel Assistant Profiles — each with its own personality, model, and RAG config.
Knowledge Base

Your files, always in context

Drop in docs and folders — surfaced in every answer.

Central-Intel Knowledge Base — local documents chunked, embedded, and searchable.

Yours to shape, yours to run

Pick any model for any job, and run it right on your own machine.

Central-Intel
Any model · one endpoint

A different mind for every job

Switch profile and model with a click — Claude, GPT, or a local model, all behind one address.

Central-Intel
Local-first

Runs on your machine

Ask anything and watch it answer on a local model — fully on-device, nothing leaves your laptop.

You choose what's exposed

Most AI tools bury this in a privacy policy. We put it on the homepage: exactly what each way of running Central-Intel reveals — and to whom.

Mode Provider sees who you are? Provider sees your prompt? Who else sees your prompt
Local / be your own serverThe model runs on your hardware
Provider sees who you are?No provider at all
Provider sees your prompt?Nothing leaves your machine
Who else sees your promptNo one
Central-Intel relayWe route to the model with our shared key
Provider sees who you are?No — no account, no IP, no ID
Provider sees your prompt?Yes — your request
Who else sees your promptOur relay — passed on, never stored
Bring your own keyYour API key, straight to the provider
Provider sees who you are?Yes — your key + IP
Provider sees your prompt?Yes — your request
Who else sees your promptNo one else
Sign in with your AI subscriptionYour Anthropic, OpenAI, or xAI account
Provider sees who you are?Yes — you logged in
Provider sees your prompt?Yes — your request
Who else sees your promptNo one else

Seeing isn't storing. On the relay, we pass your request straight to the model and never store, log, or cache its content — we only count tokens to bill you, never your messages. Honestly, we wish we didn't have to see it at all; routing the request is the single moment we touch it, and then it's gone. Want even that moment off the table? Run local.

Two different things: who you are vs. what you say. Our relay hides who you are from the provider — no account, no IP, no per-user ID. Only running local also hides what you say, because nothing ever leaves your machine. Pick the trade-off that fits the moment — and you can change it any time.

Three steps to a smarter assistant

Get up and running in minutes.

1

Install in clicks

Download Central-Intel for Mac, Windows, or Linux. It sets up local AI for you — no terminal, no config files.

2

Pick local or any model

Run a model on your own machine, bring your own key, or use ours. One OpenAI-compatible endpoint at localhost:11435.

3

Reach it from anywhere

Point any OpenAI-compatible client at Central-Intel — and with Pro, reach your home machine from your laptop or phone while you travel.

Simple, transparent plans

Start free with local models. Add relay features when you're ready. Pro unlocks everything.

Free

$0

Local models, your own API keys, profiles, RAG, and memory. No account needed.

Get Started

Basic

Pay-as-you-go

Relay access, managed credits, device sync, memory sync, and desktop tunnel.

Get Started
View Full Pricing & Compare Plans

Built for power users who value privacy

Your own intelligence. Every device you own. Loyal to you alone.

Central-Intel was built because every AI client should feel like it knows you — not just the conversation you started five minutes ago. Your notes, your calendar, your files, your memory: all of it, available to every AI tool you use, privately.

Use our hosted relay for convenience, or go fully local with llama.cpp. Run local and nothing leaves your machine at all — see exactly what each mode exposes in our Privacy Ladder.

◈ Central-Intel

# Your relay URL
OPENAI_API_BASE=
  https://api.central-intel.ai/t/your-tunnel/v1

# Works with any client
Open WebUI
Obsidian Copilot
Cursor / Continue
Any OpenAI SDK