Ask anything
The difference
Using AI today vs. using AI with Governor
You already do all of this in your head โ guessing which AI to use, wondering if a cheaper one would've gotten the job done, hoping you didn't paste something you shouldn't have. Governor just does it for you, every time.
You're the one deciding
-
1
You have something to do
Write some code, draft an email, make sense of a bill, ask a question.
-
2
You pick an AI
ChatGPT? Claude? The one running on your laptop?
Most people are guessing here โ and there's no way to check. -
3
You paste it in as-is
Your personal details go along with it. Your app doesn't check what you're sending, and the AI company receiving it has no reason to.
-
4
You get an answer
Maybe a good one. You'll never know if another AI would have done better.
-
5
Not happy? Start again
Switch apps, re-type the whole thing, try your luck a second time.
Time gone, money gone, and your details are already out there.
What it costs you
- Paying premium prices for questions that didn't need them
- Getting weak answers because the AI wasn't suited to the task
- Personal details sent out with no one checking
- Time lost switching apps and re-typing
- Starting from scratch every single time โ nothing learns
It's handled before you notice
-
1
You have something to do
Same task, same question, typed the same way. You change nothing.
-
2
Governor takes a look first
In the time it takes to press enter, and before any AI sees a thing:
- Works out what kind of help you need
- Compares the AIs you have access to
- Applies what you said matters โ cost, speed, quality, privacy
- Takes out anything personal if it's going to the cloud
- Tidies up the wording for whichever AI it picked
-
3
The right AI gets it
Quick question, quick cheap model. Hard problem, the powerful one. Point it at an AI on your own computer and that work never leaves it.
Every time, without you thinking about it. -
4
Better answer, smaller bill
The right AI with well-formed wording beats the wrong AI, and usually costs less.
And it remembers your preferences, so it keeps improving.
What you get
- Simple questions stop going to expensive models
- Hard questions go to AIs that can actually handle them
- Personal details removed, or kept entirely on your machine
- No switching apps, no re-typing, no guesswork
- It gets better the more you use it
Governor works with the AI apps you already use. Install it once, point your app at it, and every question you ask is handled automatically from then on.
Works with any app that talks to OpenAI-compatible services โ Open WebUI, LibreChat, LangChain, and more.
How it holds up
Three promises, and why they hold.
None of these is a setting to find and switch on, and none asks to be taken on trust. Each follows from how the routing decision is made, and each can be verified from inside the product.
Selection is based on measured evidence
Models are ranked per category from public head-to-head results, then scored against what your kind of task actually requires โ reasoning, speed, context length, cost. Every answer records which model was selected and on what grounds.
The unmetered models sit close to the frontier
The models included without a meter score within 25 points of the best model we can reach, on a leaderboard where the full routed roster spans 117 โ at roughly a fifth of the cost. That margin is what makes unlimited use sustainable: routing means most questions never required the expensive model, so the premium balance is preserved for the ones that do.
The model answers you without knowing who you are
Names, emails, phone numbers and addresses are redacted from your message before it is forwarded, and it reaches the provider from AI Governor rather than from you โ no account of yours, no billing identity, no address to attach a history to. Most questions are routed on extracted keywords alone; where the wording genuinely matters, the redacted sentence is what gets read. Conversations are held in your browser and nowhere else, and Settings reports exactly how much is stored and erases it in one action.
Want the details? The classification method, the learning process and the measurements behind them are set out in the paper.
Pricing
Unlimited, because most questions don't need an expensive model.
Start free, in the browser, with no account. Paid plans lift the daily cap entirely: unlimited use of the everyday models โ the best of which is within 25 points of the best model on the leaderboard โ plus a monthly balance for the premium ones, spent only when a question actually warrants it. And the desktop version is free forever, with no cap at all, because it runs on your own machine.
Free
$0 forever
Ten questions a day in the browser, or the whole thing on your own computer.
- 10 free questions a day in the browser
- The first 5 need no account at all
- Removes personal details from every prompt
- Picks the best AI for each question
- Learns your vocabulary as you go
- 300+ AI models to choose from
- Or download it and run unlimited on your own machine
First 100 people
Founding
$8 /month, locked for life
Unlimited everyday models, and a price that never goes up.
- Everything in Free
- Unlimited use of the everyday models
- $1/month toward premium models
- Smart from day one โ no learning period
- Instant signup, no trial, price locked in for as long as you stay
AI Governor Pro
$18 /month
The standard plan, with a week to try it before you're charged.
- Everything in Founding
- Unlimited use of the everyday models
- $1/month toward premium models
- Free for 7 days, cancel anytime before it ends
AI Governor Pro+
$68 /month
For people who reach for the strongest models most days.
- Everything in Pro
- $75/month toward premium models โ more than the plan costs
- Around 120 conversations on the strongest models
- Balance resets on your renewal date
AI Governor Max
$200 /month
Frontier models as the default rather than the exception.
- Everything in Pro+
- $225/month toward premium models
- Around 360 conversations on the strongest models
- Teams from $149/month โ get in touch
Your premium balance goes further here than anywhere you pick the model yourself. On a menu, people reach for the biggest model for everything; Governor reads the question first and only spends the expensive one when the task needs it. There is no API key to set up and no second bill to watch โ every reply is on us, on every plan including the free one.
Questions
The things people ask first.
Do I have to stop using ChatGPT or Claude?
No. Governor sits between you and whichever AI you already use. You keep your subscriptions and your habits โ Governor just cleans up what gets sent and, if you want, sends it somewhere better suited to the question.
Can you see my prompts?
The answer differs by version, so here is both. In the desktop app everything happens on your computer and your prompt never reaches us at all. In the browser we do relay it, because that is how it reaches the model that answers โ but it is redacted on the way through, and nothing is retained: no prompt logs, no conversation store, nothing on our side to leak, subpoena or sell. What the provider receives is a request from AI Governor with the identifying details already stripped, which is a different thing from a request from you.
Won't the AI give worse answers without context?
Your full question still reaches the AI that answers it. What Governor keeps private is the separate step of working out what kind of question it is, which only needs a few words. And because it picks a better-suited model, answers often improve.
Is it hard to set up?
No. Install it, point it at the AI accounts you already have, and carry on as normal. There's nothing to configure unless you want to.
What does it actually cost me to run?
The desktop app is free and runs on your own machine. In the browser, we pay for the replies โ there is no API key to set up and no second bill to watch. Either way the routing is what keeps the cost down, because simple questions stop being sent to expensive models.
What if I don't want anything sent at all?
Point Governor at an AI running on your own machine and it'll route everything there instead, so nothing leaves your computer at any point.