If you are responsible for how AI is used in a regulated or safety critical organisation, you already know the problem. It is not that there is too little guidance. It is that there is far too much of it, arriving from too many directions, changing too often, and almost none of it written with your Tuesday afternoon in mind.

Keel is our answer to that. It is a weekly briefing, checked by a person before it reaches you, that tells you what changed in AI regulation and standards this week, who it affects, and what you might need to do about it. Nothing more, and deliberately nothing less.

Here is why we built it.

The people carrying the risk are the least supported

Over the last two years AI has moved from pilot projects to the working day. Clinical teams use it to draft notes. Compliance teams use it to screen transactions. Legal teams use it to review contracts. Most of this is good, and most of it is happening whether or not anyone signed it off.

Somebody in each of those organisations is accountable for that being done properly. A Data Protection Officer. A Senior Information Risk Owner. A Clinical Safety Officer. A Head of Risk, a Compliance Director, a General Counsel. The job title varies, but the position is the same: you are the person who has to be able to say, with a straight face, that the organisation understands the AI it is using and is meeting its obligations.

That was a hard job before AI. It is now a job where the rulebook is being rewritten while you are trying to follow it.

The rules are not settled, and they are not coming from one place

In the UK alone, guidance on AI is being produced by the ICO, the MHRA, the FCA, the CQC, NHS bodies, the Department for Science, Innovation and Technology, and a growing list of standards organisations. Add the EU AI Act, which reaches anyone serving European customers or patients, and international standards like ISO 42001, and you have well over a dozen sources that each expect you to have read them.

None of them coordinate. Each publishes on its own schedule. A consultation closes here, a new code of practice appears there, an enforcement decision quietly sets a precedent somewhere else. Individually, most of these updates are small. Collectively, missing them is how organisations end up discovering their obligations in the least pleasant way possible.

The honest reality for most governance leads is that keeping up properly would take a day a week. Nobody has that day. So the job gets done in snatched moments, from newsletters and LinkedIn posts, with the nagging feeling that something important slipped through.

Why this matters beyond the compliance team

It would be easy to frame this as a paperwork problem. It is not.

In healthcare, an AI tool that has not been assessed properly can miss the patient it was supposed to flag. In financial services, a model nobody fully understands can quietly disadvantage a whole group of customers. In legal and professional services, a confident but wrong output can end up in front of a court or a client. The rules exist because the consequences are real, and the people trying to follow the rules are often the last line between a promising tool and a serious incident.

Getting AI right is not a brake on progress. It is what allows an organisation to adopt these tools with confidence rather than crossing its fingers. But that confidence depends on knowing where the ground is, and the ground keeps moving.

What Keel does

Keel watches the sources so you do not have to.

Every week we scan more than a dozen regulators, standards bodies and official channels for anything relevant to how AI is governed, procured, assured and deployed. Most of what appears is noise. Some of it matters a great deal. Keel's job is to tell the difference and hand you only the part that matters, in plain English, in a briefing that takes around five minutes to read.

Each item answers three questions:

1. What changed? The actual development, stated plainly.
2. Who does it affect? Which sectors, roles and types of organisation should pay attention.
3. What might you need to do? A practical read on whether this is something to note, something to raise, or something to act on.

No jargon. No twenty page PDF you will never open. No pretending that a minor consultation is a crisis, or that a genuine shift is business as usual.

Every briefing is reviewed by a person before it is sent

This is the part we want to be completely clear about.

Keel uses automation to do the heavy lifting of monitoring sources, because no team can read everything every week. But automation on its own is not good enough for this job. Regulatory language is subtle. Relevance depends on context. A system that summarises confidently but misjudges what matters is worse than no system at all, because it makes you feel covered when you are not.

So nothing reaches your inbox without a member of our team reading it first. We check that what has been flagged is accurate. We check that the assessment of who it affects is right. We cut what does not deserve your attention and we sharpen what does. If we are unsure about something, we say so rather than dress it up.

We build AI assurance systems for a living. We know exactly where these tools are strong and where they fall over. Keel is built on that knowledge, and the human review is not a disclaimer. It is the product.

Who it is for

Keel is for anyone in a regulated or safety critical industry who is accountable for AI being used well inside their organisation. Our first readers are in health and care, financial services and legal, but the principle applies anywhere the cost of getting AI wrong is measured in more than money.

If you have ever forwarded yourself a regulator's press release with the intention of reading it properly later, Keel is for you.

Where we are

We have just sent our first briefing to a small group of people whose judgement we trust, and we are asking them to be blunt. Over the coming weeks we will be refining what Keel covers, how it reads and how it is delivered, based entirely on what those readers tell us.

If you would like to be one of them, or you simply want to see a briefing for yourself, get in touch. We would rather build this with the people who need it than guess at what they want.

Join Keel now: https://keelwatch.co.uk/