Honest Answers

The questions you should be asking, answered straight.

You have heard a hundred AI pitches, and you are right to be sceptical of this one too. Confidential work does not leave room for taking a vendor's word for it. So here are the questions a careful professional actually asks, answered plainly, with the proof you can check yourself. No spin, no hand-waving.

"Is a private model actually good enough for real work?"

For the work most professionals need it for, yes. Drafting letters and emails, summarising a long document, turning an inbound report into a clear checklist: today's private models are already at the level.

Frontier models pull ahead on the very hardest reasoning and maths problems, not on a referral letter or a client update. And because you review every output before it goes anywhere, you are never trading safety for quality. You get both. The honest test is not our word, it is your own work: try it free, on the real thing, and judge the output yourself before you pay a penny.

Check it yourself: run a real task through the live demo on the home page, no account needed. Compare the three models on the models page, or see how Hush stacks up against other providers on the comparison pages.

"You are early-stage. What happens to my work if you disappear?"

Nothing of yours is trapped, because nothing of yours is held. Your chat history lives encrypted on your own device, not on our servers, and you can export or wipe it in one click.

There is no proprietary format to be locked into, no data held hostage, no per-seat contract to unwind. Billing is month to month. If Hush vanished tomorrow, you would lose a tool, not your files and not your clients' confidences. The very thing that makes us private, keeping no conversations or documents, is what makes us safe to try: there is nothing of yours for anyone to strand.

Check it yourself: how your data is handled and what we do and do not keep.

"Why should I trust your claims when you have no big-name customers?"

Do not trust us. Verify us. That is the whole point, and it is why the proof is built to be checked rather than admired. The verify-us page puts the registers and the copy-paste terminal checks in one place.

The company is verifiable today at Companies House (17278687) and the ICO (ZC126901). Our AI sub-processor list is public, and it is empty. Every plan includes a one-click audit-log export you can inspect. And the free pilot runs on your own real work, so the evidence is yours, not another firm's testimonial. One proof point you can check on your own matter is worth more than ten logos you cannot.

You do not have to take our word that the risk is real, either. A Microsoft lawyer told the French Senate, under oath, that he could not guarantee data would never be handed to US authorities. A UK tribunal has confirmed that putting confidential material into public AI can place it in the public domain and waive privilege. The danger is on the record. So is our answer to it.

Check it yourself: what was said at the French Senate and the privilege decision, for solicitors.

"Is £39 a month too good to be true? Where is the catch?"

There is no catch, and the reason is structural. Because we own the hardware that runs your work, there is no Big Tech API bill sitting underneath us to mark up.

So £39 reflects what it actually costs us, not a loss-leader designed to hook you and raise the price once you depend on it. No card to try it. No auto-renewal ambush. No per-seat gouging as your team grows. And set it against the real number: for a small firm, the cost of a single confidentiality breach, the ICO exposure, the lost client, the professional-indemnity fallout, dwarfs a year of Hush. It is priced deliberately so that a solo practitioner is never priced out of privacy.

Check it yourself: the full pricing, in plain numbers, and the time-savings calculator.

"Isn't your own hardware slower or less reliable than Big Tech's cloud?"

For the work you actually do, there is no difference you would feel. The Prism model is tuned for near-instant responses, and owning the hardware is exactly what lets the privacy promise be real rather than a policy that could quietly change.

We will be straight about the trade you are making: we are a smaller team, so when something on our side needs attention, we tell you plainly that we are updating and to try again shortly. We never dress it up as something else. What you get in return is privacy with no penalty you would notice in a working day.

Check it yourself: try the live demo and time a real response.

"Will my DPO or IG lead actually approve this?"

It is built for exactly that conversation. Hush is an ICO-registered data controller, runs on hardware we own with no US parent and no third-party AI sub-processor for your content, and gives you a DPA draft and a DPIA template on request alongside a one-click audit-log export.

Hand your DPO the free pilot and let them test the claims against your own workflow, on your own terms. That is the strongest position you can be in: you are not asking them to trust a brochure, you are giving them something they can verify. You bring proof, not a promise.

Check it yourself: the procurement and IG pack, the security detail, and how AI use affects your PI cover.

The fastest way to answer every question above is to use it on your own work.

Free to try right now with no account. Then two weeks free on every model and tool, no card, no commitment, no conversation or document of yours kept. Share it and get two more.

Try Hush free →

Your own work, on your own terms, with evidence you can verify.

Have a question that is not here? Email [email protected] and we will answer it straight. Or read the full FAQ.

Hush AI is a drafting and summarisation assistant. It is not a source of medical, legal, or financial advice. Always verify AI-generated outputs before acting on them.

Written and maintained by the founder, Dr W.J Carter. We would rather you verify us than take our word for it.