Trust & security centre

How processing actually works

A tool on this platform runs in one of two places, and every tool page states which before you give it anything.

In your browser (82 tools)

PDF, image, data and text tools run as code inside your own browser tab, using WebAssembly and the browser's built-in graphics and crypto engines. Your file is read from disk by the page you already have open. It is never transmitted. The network request that loads the tool is the last network request involved.

How to verify it: open your browser's developer tools, go to the Network tab, then run any tool marked Stays on your device. You will see the page and its code load, and then nothing further. No request carries your file, because no such request exists in the code.

On our servers (AI tools)

Language models are too large to run on your device, so AI tools send your text to a server, which forwards it to an AI provider. This is a real privacy trade-off and we describe it as one:

• The provider that receives your text is named on the tool page, above the input box, before you type anything.
• Your input is held only for as long as the request takes, then discarded. It is not written to a database.
• Your content is not used to train any model, and our provider agreements must prohibit it. If a provider changed that, we would drop the provider.
• We do not keep a copy to "improve the service".

If a document is sensitive enough that this matters, use a tool that runs in your browser instead. We would rather tell you that than take the upload.

What we store

Without an account, we store nothing that identifies you. Specifically:

• No file contents. For in-browser tools there is nothing to store. For AI tools, nothing is retained after the response.
• No file names. They are often descriptive of private matters, so they never leave your device even in an analytics event.
• No tracking cookie and no advertising identifier. There is no advertising on this platform.
• No successful search queries. Search runs in your browser. Only searches that return nothing are logged, as the query text alone, so we can teach search the words real people use.

What we do count, in aggregate and unlinked to any person:

• Which tool was opened, started, completed or failed, and how long it took.
• Whether the device was a phone, tablet or desktop — not its exact screen size.
• The country the request came from, derived from the request at the edge.
• Thumbs up or down, plus an optional reason you chose from a list.

We honour the browser Do Not Track setting. With it on, no analytics event is sent at all.

If you create an account

Accounts are optional and no tool on this platform requires one. If you make one, it exists to remember your favourites and recent activity. Then we hold your email address, the activity you chose to save, and your settings. You can export all of it, and you can delete your account, which deletes the data rather than deactivating a flag on it.

Security practices

• Everything is served over HTTPS with HSTS, and HTTP is redirected.
• A strict Content Security Policy is applied; no third-party analytics or advertising scripts are loaded.
• Database access uses row-level security, so a row can only be read by the account that owns it — enforced by the database, not by application code that might have a bug.
• Privileged service credentials exist only in server-side environment variables and are never present in anything sent to a browser.
• Uploaded files, where a tool needs them, go to a private bucket reached only through short-lived signed URLs, and are deleted automatically by a scheduled job.
• Rate limits apply per IP and per account to keep abuse from degrading the service for everyone.
• Sensitive administrative actions are written to an append-only audit log.

What we have not certified

This section exists because its absence elsewhere is usually the tell.

We hold no security certifications. There is no SOC 2 report, no ISO 27001 certificate and no HIPAA attestation for this platform, and we will not display a badge implying otherwise. Those things require an external audit, and we have not had one.

We have built to the principles those standards test for — least privilege, encryption in transit, data minimisation, auditability. That is not the same as being certified, and we will not blur the two.

On data protection law: because in-browser tools collect no personal data at all, most of the surface area that regulation covers simply does not exist here. Where we do process personal data — an account, an AI request — we apply the rights those frameworks establish: access, export, correction and deletion. If you need a formal data processing agreement for organisational use, contact us rather than assuming one is in place.

AI data handling

• An AI tool tells you which provider will receive your text before you submit it.
• Your content is not used to train models.
• Prompts and responses are not stored after the request completes.
• Model output can be wrong, and we say so on the tools where being wrong would matter — contracts, policies, anything with legal or financial consequence.
• We do not present AI output as professional advice, and we do not let a tool imply it was reviewed by a person when it was not.

Data retention

• In-browser tool files: never received, so nothing to retain. They exist in your tab's memory and disappear when you close it.
• Temporary uploads (only for tools that require a server): deleted by a scheduled job within 60 minutes, and immediately once the job finishes.
• AI prompts and responses: not persisted.
• Aggregate analytics: retained as counts, with no identifier to tie them to a person.
• Account data: kept until you delete your account, then removed.

Reporting a vulnerability

If you find a security problem, email naseershafqat@gmail.com with enough detail to reproduce it. We will acknowledge within three working days and tell you our assessment and intended fix.

We will not pursue legal action against anyone who reports a vulnerability in good faith, gives us reasonable time to fix it, and does not access or alter other people's data in the process. We do not currently pay bounties, and we say that upfront rather than leaving it ambiguous.

If something goes wrong

If we suffer a security incident affecting personal data, we will publish what happened, what data was involved, what we did about it, and what we changed — within 72 hours of establishing the facts, and notify affected account holders directly. We will not describe an incident as "a technical issue" if it was a breach.