RunAgentRun.
Policy

Privacy & Cookies

The short version: we set no cookies at all, we run no ads, and we never sell data. Our analytics are cookieless and anonymous. Here is the whole picture, in plain English.

Who we are

RunAgentRun (runagentrun.co.uk) is an editorial publication about practical agentic AI and automation for UK small businesses. For anything in this policy, contact us via the details on the About page.

Cookies

This site sets no cookies at all — none for analytics, none for tracking, none "essential". That is why you see no cookie banner: there is nothing to consent to. The only way a cookie could appear is if you choose to load an embedded post from X (Twitter) on an article — these are click-to-load, so nothing from X runs unless you press the button, and X's own cookies are theirs, not ours.

Analytics — what we collect and why

We use PostHog (hosted in the EU) in a deliberately cookieless, anonymous mode: it counts page views and aggregate interaction signals without storing anything on your device and without building a profile of you. We use this only to rank "Most Read", improve the site's structure, and decide what to write next.

  • No cookies, no device identifiers, no cross-site tracking, no advertising.
  • No personal profiles — every visit is anonymous; we never identify you.
  • We do not sell or share data with anyone.

To count unique visitors without cookies, we use the same approach as privacy-first analytics like Plausible and Fathom: when a page loads, our server combines your IP address and browser type with a secret that changes every day, and keeps only a one-way hash of the result. That hash lets us tell two page views apart on the same day, but it cannot be reversed to your IP and cannot link you across days — so it's a counter, not a tracker. Your raw IP address is never stored: it's used once for an approximate country lookup, then discarded.

The only thing we keep in your browser is a single optional flag — set only if the editor visits with ?internal to exclude their own visits from the counts. Readers never get it.

Newsletter & RSS

Our RSS feed requires nothing from you. If you subscribe to the email digest, your address is handled by Buttondown (our email provider) solely to send you the newsletter, with one-click unsubscribe in every email. We don't use it for anything else, and we don't sell or share it.

Rover — the email research assistant

Rover is a separate service from the website. You start it by emailing it; it emails you back with what it has found about the things you asked it to watch. Because that involves your email address and your own words, it is the one part of what we run that handles real personal data — so this section is specific rather than general.

What Rover holds about you

  • Your email address, and the address any reply came from.
  • The text of the emails you send it — including the questions you ask and anything you forward to it — and the subjects and briefs you asked it to follow: the company or person, the angle, and anything you told it you were not interested in.
  • The briefings it sent you, kept so it does not repeat itself and so there is a record of what you were actually sent.
  • The first name it greets you by, and — only if you tell it — what you do and what you will use the briefings for. It uses these to say why a story matters to you, reads them back to you when you give them, and never adds to them from your email address, your signature or anywhere else.
  • Which "more like this" / "less like this" links you pressed, if you pressed any in older briefings (new briefings no longer carry them; you steer by replying).
  • If you turn on the spoken briefing: the recording itself, a count of how many times it was requested, and when it was last requested. Nothing about how much of it you listened to — we could not know that, and would not want to.
  • The numbered list of stories in each briefing, so that a reply like "more like the second one" can be tied to the story you meant. What you say about a story in a reply is kept as a preference, the same way a "more / less" click was.
  • If you ask it to watch a web page: the address, and the public text of that page as it last read, so it can tell you what changed.
  • The relevant findings Rover verified for you — including ones that didn't make the cut for space — so it can spot patterns across them. Each is a public fact with its source; kept for 60 days. And which subjects it has suggested changes for, so it doesn't ask twice in a week.
  • The layout choices you make with /settings — which parts of the briefing to show and how many stories.
  • If you ask for alerts or a second reader (for example "what it means for our compliance team"): those settings, in your words, and when the last alert was sent.
  • A browser copy of each briefing, so the "view in your browser" link works. Anyone you give that link to can read that one briefing; it is not listed or indexed anywhere, and it is deleted after 30 days.
  • Timestamps, and counters that enforce the daily limits.

Your file on a subject (the link beside each story) is not extra data: it is a page built from the stories Rover has already sent you. Like the browser copy, anyone holding the link can read it, so share it only with people you'd forward the briefings to.

Rover sets no tracking pixels and no open-tracking. It does not know whether you opened an email, only what you deliberately clicked.

The spoken briefing

The ninety-second recording is off unless you turn it on, from a link at the foot of any briefing, and you can turn it off again from the same place. When it is on, the script — written from the same verified findings — is sent to MiniMax to be read aloud, and the resulting audio file is stored on our server in Germany alongside the briefing it belongs to. Its web address is a random 24-character string that is never listed or indexed, and it is deleted on the same 90-day clock as the written briefing — the file itself, not just the record of it. If you delete your data, the recordings go with it.

Why we are allowed to hold it

Consent, under Article 6(1)(a) UK GDPR and the UK's PECR rules on electronic mail: you started the conversation by emailing us, and the briefings exist only to do the thing you asked for. You can withdraw that consent at any time — every briefing carries an unsubscribe link, your mail client's own unsubscribe button works, and replying STOP does the same. Withdrawing is as easy as starting.

Who else sees it

We use a small number of processors. None of them is paid with your data and none of them is an advertiser.

  • Hetzner (Germany) — the server everything runs on. Your data is stored here, in the EU.
  • Resend (United States) — sends and receives the email itself, so it handles your address and the message body.
  • MiniMax (outside the UK and EEA) — the language model that reads the sources and writes the briefing. It receives the names and descriptions of what you asked us to watch, your first name and anything you told Rover about yourself, what the briefings are for and any second reader you named, any question you ask Rover, the covering note on anything you forward it, and the text of public web pages we fetched. It does not receive your email address.
  • Serper (United States) and Supadata — web search and video transcripts. They receive a search term or a public URL, never anything about you.

The model provider is outside the UK and EEA, so that part is an international transfer. We limit it to the smallest thing that works — your interests, never your identity — and we do not send it your address, your name or your message history.

We do not train anything on you

We do not use your data to train or fine-tune any model, and we never will without asking you first. What the model provider does with the text it receives is governed by their own terms, which is exactly why we send them as little as we do.

How long it is kept

These are the periods the system actually enforces, not an aspiration — the deletion runs every day:

  • Briefings sent and emails received — 90 days
  • Spoken recordings, and their play counts — 90 days
  • The record of which stories you have been told, so it does not repeat itself — 180 days
  • Your "more / less" preferences, from clicks or replies — 365 days
  • The numbered list of stories in each briefing — 90 days
  • Browser copies of briefings — 30 days
  • Verified findings kept for spotting patterns — 60 days
  • Watched pages, alert and second-reader settings — until you remove them, unsubscribe or delete your data
  • Your account and what you asked it to watch — until you unsubscribe or delete it

Your rights, and the button

You have the usual UK GDPR rights: access, rectification, erasure, restriction, portability, and objection. For Rover most of them are a link rather than a request — the unsubscribe page in every briefing has a button that deletes everything: your address, your briefs, every briefing sent to you, and every click you made. It is immediate and cannot be undone. Anything else, reply to a briefing or use the contact on the About page, and we will answer within one month. If you are not satisfied you can complain to the ICO.

Decisions, and what Rover is not

Rover summarises public information. It makes no automated decision that has a legal or similarly significant effect on anybody, so Article 22 does not come into it — and nothing it writes is advice. Every briefing says on its face that it was written by an AI assistant, and every quote in it links to the source it came from so you can check it yourself.

AI-generated content

Articles here are drafted by an AI agent and reviewed by a human editor before publication — a process that's itself on the record at Inside the Newsroom. Reading the site involves no AI processing of your data; this note is for transparency about how the words are made. What the content is and isn't is covered in our disclaimer, and the rules of use in our terms.

Article images

Photography is sourced under licences that permit commercial use (e.g. the Pexels licence or Creative Commons); attribution is shown beneath article images where the licence calls for it.

Your rights

Under UK GDPR you can request access to, correction of, or deletion of any personal data we hold. Since our analytics are anonymous and we set no cookies, the usual answer is "there is nothing to delete" — but ask any time via the About page.

No ads. No data sales. No cookies. That's the policy.