Privacy Policy

1. Who is responsible for your data

The data controller is Oleksii Vaschenko, an individual based in Norway. There is no registered company behind Extruse, so there is no registration number to give. If one is registered, its name, address and number will appear here.

Reach us at hello@extruse.com about anything on this page.

2. Using Extruse without an account

The whole calculator works without signing up. In that mode no server is involved in your data:

• Your workshop — printers, materials, products, spools, print queue, print log and history — is stored on your own device.
• G-code, 3MF, STL and OBJ files you open are read inside your browser or app and are never uploaded. Only the numbers we extract from them, such as print time and filament weight, are kept, and only on your device.
• A quote you share as a link carries the quote inside the URL fragment — the part after #. By design a fragment is never sent to any server, ours included.
• Clearing your browser storage, or deleting the app, erases this data permanently. We cannot restore it, because we never had it.

One thing does reach a server even without an account, and it is described in section 9: your address is looked up to guess a country, so the site opens in a language you are likely to read.

3. Using Extruse with an account

If you create an account, we store:

• your email address and display name, to identify the account;
• your workshop data — printers, materials, products, spools, stock movements, print queue, print log, clients, quotes and settings — so that it reaches your other devices and survives a lost phone;
• subscription status, once paid plans exist.

This is held in our database at Supabase, hosted inside the European Union. Access is enforced at the database level: rows are readable only by the account that owns them. Model files are still not uploaded — the account changes what happens to your numbers, not to your files.

There is one exception to that rule, and it is us. A small support console lets us see your email address, your plan, how many records you have, when your last sync was, which paid features you ran into, and any note we have written about a conversation with you. It does not show the contents of your workshop — not your prices, not your clients, not your history. Exporting your workshop in full is possible, but only by the owner of the service, only with a stated reason, and every such export is written to a log that cannot be edited or deleted.

We do not sell your data, and we do not use it to train models — not your prices, not your customers, not your production history. Neither does anyone we send data to. Your workshop is your business, not our product.

We do count things across the service as a whole: how many workshops set a failure rate above ten per cent, which presets get picked, where people abandon a form part-way. Those counts carry no name, no customer and no workshop, and they are how the tool gets better. If that ever changes — if we want to use what you record for something else — this page will say so before it starts, not after.

4. Your customers’ data

Extruse has a clients section. What you put there — a name, a phone number or messenger handle, an email address — is personal data about someone who never agreed to anything with us. That changes the roles between us, and it is worth stating plainly.

You decide what to record about your customers and why: in the language of the GDPR you are the controller of that data. We only hold it so that it reaches your other devices and survives a lost phone: we are your processor, and we act on your instructions, which in practice means storing it and giving it back to you.

Concretely, and for as long as this page says so:

• we do not read your clients, mine them, or use them for anything of ours;
• we do not share them with anyone, and we never sell them;
• they live in the same European database as the rest of your workshop, under the same row-level rules, and are covered by the same support console limits in section 3 — which is to say the console does not show them;
• when you delete a client, or your account, their data goes with it on the timetable in section 13;
• if we ever need a sub-processor beyond the hosting already named on this page, it is named here before it starts.

What stays with you: having a lawful reason to hold your customers’ details in the first place, telling them you do, and answering them if they ask. We cannot do any of that for you, because we do not know them — and by design, we do not look.

5. Signing in with Google

If you choose Google sign-in, we ask Google only for your basic profile and email address. We never see your Google password. You can revoke our access at any time in your Google account permissions; that stops future sign-ins but does not by itself delete data already stored — use section 12 for that.

Email-and-password sign-in is available if you prefer not to involve Google.

6. Email we send

Account email — confirmation, password reset, and notices about a quote your client reacted to — is delivered through an email provider on our behalf. They receive the address and the message, and nothing else about you.

7. Payments

When paid plans exist, payments will be processed by a payment provider. Card numbers are entered on their systems and never reach our servers; we receive only the outcome of a payment and what an invoice needs.

8. The assistant

Extruse can offer to do twelve things with a language model: fill in your workshop from a description you type or speak, read a photo of a spool label or a page from your notebook, draft a marketplace listing, tell you what similar items usually sell for, put your own cost calculation into words, read a price list you paste or photograph so you can see where your prices stand next to it, turn a customer’s message into an order draft, read the weight and print time off a screenshot of your slicer, put your monthly report into words, draft replies to a customer about a quote you sent, and turn a spoken or typed phrase into a print-log or maintenance entry, and read a search phrase so the app can find your own record.

All twelve are optional. Every screen that offers one has a manual path that does the same job, and the assistant never saves anything by itself — it proposes, and you confirm. It also never does the arithmetic: every number you see next to its words is computed by the app from your own data, and a reply in which the model wrote a number of its own is discarded.

What leaves your device, and only when you ask for it. For the workshop: the words you wrote or spoke, or the photo you took, together with the names in our printer and material catalogue so the model can recognise “Bambu A1”. For a listing or a price: the product’s name, its price, its unit cost, how long it prints, which marketplace you picked, and the product’s photo if you have added one. For an explanation of a calculation: the cost lines of that one calculation — their names, amounts and shares — and nothing about the product or the printer. For a price list: the text or photo you provided, plus the names of your products so the model can match them — not their prices and not their costs; the comparison is done on your device. For an order from a message: the message you pasted, today’s date, and the names of your clients and products so the model can match them — not their contacts, not their order history, and not your prices; the price is yours to set. For a slicer screenshot: the screenshot alone; the material and printer it names are matched to our catalogue on your device. For the month in words: the month’s totals (sales, revenue, profit, printer hours, successful and failed prints), the names of your best-selling products with their quantity and profit, and the same totals for the previous month; the model writes no numbers, and the figures shown next to its words come from your own report. For a reply to a customer: that one quote as you sent it (the customer’s name as you saved it, the item, price, quantity, lead time, validity date, payment ways, the customer’s note and the quote’s status) and a floor price that the calculator worked out before the call; the model may only repeat the numbers it was given, and a draft with any other number is discarded before you see it. For a log entry in words: the phrase you said or typed, which of the two forms it is for, and — for a print entry only — the names of your printers; nothing is saved until you confirm the fields yourself. For a search in words: the phrase you typed and today’s date, and nothing else — your records never leave the device for this one; the model returns which kinds of record and which period to look at, and the search itself runs on your device. Nothing else goes with any of these — not your clients’ details, not your sales history, not your other products.

Who runs the model. Anthropic, a company in the United States, and the request is processed there: for as long as the answer takes, what you sent is outside the European Union. They are our processor for this and nothing else. Their commercial terms forbid them to train models on what we send, and neither do we. If we ever change provider, this page will name the new one before the change, not after.

Speaking is not sent anywhere. The microphone uses your own device’s speech recognition, and it appears only when that recognition runs on the device itself. If your device does not yet have the speech model for your language, the app offers to download it once — onto the device, not into a service. Your voice reaches neither us nor the model; what we receive is the text your device produced from it.

What we keep. One line per request: when it was, which task, which model, how many tokens and what it cost us. We do not keep what you asked or what came back, and there is nowhere to keep them — the table has no column for text.

When it is off. The assistant can be switched off, it can run out of its monthly allowance, and it can be unavailable. In every one of those cases the buttons simply do not appear, and everything on those screens works exactly as it did before the assistant existed.

9. What our servers see

No access log. Our web server is configured without one, so there is no file recording which address asked for which page and when. What it does write is error output — a fault, a stack trace — which does not identify visitors.

Your country, to pick a language. The site speaks ten languages, and the first time you arrive we have to guess which one. Your browser’s language setting is a poor guide on its own — plenty of people run an English browser and do not read English — so we also ask a lookup service, api.country.is, which country an address belongs to. Your IP address is sent there and the two-letter answer comes back; the answer is kept in a cookie for a day so the question is not asked again on every page. We do not store your IP address for this, and the service is not ours.

Your sessions. When you are signed in, our authentication provider records each session with the address it was created from, the browser that created it and when it was last active. That exists so you can look at the device list in your profile, recognise your own sessions and close one you do not recognise. The country shown beside each session comes from the same lookup described above.

10. When you send us a support snapshot

The app has a “something is wrong” button. It does nothing until you press it, and pressing it sends us, in one go:

• what you wrote in the box;
• up to three screenshots, if you attached them — these are pictures of your screen, so look at what is on it before you attach one;
• what the app was running on: platform, language, theme, screen size and build;
• the state it was in: your plan, currency, whether sync succeeded or failed and the last errors it recorded;
• how much you have, as counts — how many printers, materials, spools, products. Counts, not contents.

It is sent to us as an email so that a person can answer it. Nothing is taken without that button being pressed, and everything above is deleted after 30 days (section 13).

11. Tracking, and what we do count

Extruse contains no advertising and no third-party analytics. There is no tracking cookie to consent to, because nothing here follows you between sites.

Two things about your use of the product are recorded, and both are ours alone:

• when you were last active, as a timestamp on your account, so we can see whether anyone is using the thing at all;
• which paid feature you ran into and when. Not what you were doing — only that a locked door was reached, and which one. It exists so that what gets built next is decided by where people actually stop, and it is the reason a limit ever gets raised.

Three cookies, all necessary, none for advertising:

• a session token, if you are signed in, so that you stay signed in;
• your language choice, kept for a year, so we stop guessing once you have said;
• the country answer from section 9, kept for a day, so we stop asking.

We may add crash reporting, so that faults get fixed instead of guessed at. If we do, this page is updated before it starts, and what it collects is named here.

12. Your rights

If you are in the EU or UK, the GDPR gives you the right to:

• see what we hold about you;
• take it with you — Extruse has a one-click export that writes your entire workshop to a file, so you do not need to ask us;
• correct anything wrong;
• delete your account and its data;
• object to processing, or complain to your national data protection authority.

Write to hello@extruse.com and we will act within 30 days. Our legal basis is performing the contract you enter into by using the service, and our legitimate interest in keeping it secure and working.

13. How long we keep things

Account data is kept while the account exists. Deleting your account starts a seven-day pause — enough to undo it from the link we email you — after which the account and its workshop data are removed from our live database. Backups are taken daily and kept for seven days, so a deleted account is gone from those within a week as they age out.

Two things outlive the record they belong to, and both for the same reason — your other devices have to learn that something was deleted. A deleted workshop record leaves a marker (its identifier and the time, not its contents) for 180 days, after which the marker goes too. A support snapshot you send us from the app is kept for 30 days and then removed with its screenshots.

Invoicing records, if there ever are any, are kept as long as tax law requires.

14. Children

Extruse is a tool for running a business and is not intended for anyone under 16. We do not knowingly collect data from children.

15. Changes to this policy

If we change how we handle your data, we update the date at the top and, for anything material, tell account holders by email before it takes effect.