for agents/llms.txtv0.14.203

Home / Articles / The way in: why sovereign websites matter so much to RiskMandate, what Web Summit validated and what it showed was missing, something to buy now, how every technology I have been building meets in one product, and why nobody deploys agents they have not touched, seen and been able to stop, so the personal customer becomes the way to the corporate one

The way in: why sovereign websites matter so much to RiskMandate, what Web Summit validated and what it showed was missing, something to buy now, how every technology I have been building meets in one product, and why nobody deploys agents they have not touched, seen and been able to stop, so the personal customer becomes the way to the corporate one

By · 2026-11-21 · article v1.0.0 · newsroom v0.14.189 · riskmandatesovereign-websitestrategybehaviour-policiesagentic-cellsgo-to-marketarticle

Abstract: Web Summit gave us strong validation of the technology, the direction and the need: people understood the principle, saw it working and talked to people who had used it, and even the larger offerings drew real interest. What it also showed is that nobody came to us saying "I'll buy one of these now". Not because they did not see the value, but because they could not see where to start: no thing to buy today, for a known price, and be using tomorrow. RiskMandate is learnt by using it; the summit team had its momentum because Vicky Wang ran a full newsroom without writing a line of code, delegating to a team kept under behaviour policies. Sovereign websites are the answer to where to start. A site is something everyone needs, bought one-off, cheap to make and to run, with no lock-in, version control, provenance and Claude or ChatGPT as the editor, and every one is made and run by agentic cells under RiskMandate's policies. Using it is experiencing those policies: start, stop, delegate, visualise. My hypothesis is that nobody deploys technology they have not tried, trusted and been able to control. B to C to B: anyone who could be a corporate customer can first be a personal one.

The way in: buy now, experience it, see and control it, trust it, then bring it to work.

In short

What the summit validated

At Web Summit we had validation of the technology, the path and the direction. Validation that there is a need. Validation that people understand the concept and the principle: they saw it in action, in the X-ray and the Galaxy, and they talked to people who had experienced it. The energy was positive. Even the price points held: no one recoiled, and even the main offerings, up to the £20,000 tier, drew good interest (partners sell, we build).

What it showed was missing

That is also an important signal. The people we met did not come to us and say: great, I will buy one of these now. Not because they did not see the value, or not mainly. They did not see where to start. There was no thing you could buy and use right now, no "I spend this many pounds, I get this, and I have it tomorrow". The larger offerings have a sales cycle, and that is fine, but we need a fast sale and a fast deployment as well.

Why using it is the point

So much of RiskMandate is learnt by using it: you start, you execute, you get your head around it. The summit team had its momentum not only because of what we made, which was spectacular, but because of how it was made, how fast, and how deep. We built a full news publication (PrompTugal.news): 73 articles in ten days, every one signed. And that gave Vicky the experience of running a newsroom without ever writing a line of code, as a project manager. She did not need to code. She delegated, to a development team that was controlled by behaviour policies.

That is the part that matters. It was possible because every agent was kept under control. I do a great deal of work this way now, and because the agents are limited in what they can do, I can delegate far more important tasks to them. Claude is an extraordinarily capable tool: at the same time a bicycle, a car, a Formula One car, a tank, a boat and probably a plane. You have to use the right capability for the job. By zooming in on the role, the function and the behaviour, we have a strong way of controlling what agents do (feature behaviour policies). The hard part is how to start.

From cells to websites

That is why we started selling agentic teams: AgentiCCells, a cell of agents run under RiskMandate's policies. Then the question was what these cells can do. One of the things they can do is make our websites. And once I had wired in the automation, the GitHub repository, the address in AWS Route 53 and the rest of the settings, I realised we had a new product: websites made easily, with the identity that connects to them (the website cell). There are still many details to sort out, as the first round trip showed, but for the first time we have something people can try now and buy now.

Every technology meets here

When I mapped its properties, I realised how strong this offering is. It ticks the boxes I always talk about for a startup: a one-off service, cheap to make and cheap to run (sovereign websites, the investment case). And it is where everything I have been building meets:

It also puts me right in the conversation I have been having for years: sovereignty, determinism, provenance.

Nobody deploys what they have not touched

To buy a product, people have to trust it, and to trust it they have to try it. Nobody deploys technology in their organisation that they do not understand, do not control and cannot see. That is one of the problems with agents: they are powerful, but if you cannot see what they do, you will not deploy them, even if they are fine. You need the stop button, and you need to see it. What we had at the summit that was special was exactly that: start, stop, control, delegate, visualise and map (an agentic console).

So the hypothesis is simple. Someone will not buy or use this until they have experienced it, tried it and gained trust in it, until they say: yes, this works, I like it, I am going to use it for other things. A sovereign website is where that happens. The owner talks to the agents, sees their work, and sees what they are and are not allowed to do.

B to C to B

Every person needs their own website and their own brand. You do not wait until you are looking for a job to build your brand. Everyone needs their own newsroom, to share what they know, and this is the way to do it. Which means that anyone who could be a corporate customer of RiskMandate can first be a personal customer of a sovereign website. B to C to B: the consumer product that people try now and buy now, which is critical for every startup, and the path from it into the company. A profitable way for people to learn, try and experience RiskMandate's policies, on something they all need.

What this does not claim

Where this comes from

My debrief in writer-10's session, on 21 November 2026. Figures on PrompTugal.news are from its own article. Related: the first round trip, the website cell, starting AgentiCCells, PrompTugal.news, the newsroom we ran, feature behaviour policies, an agentic console, sovereign websites, the investment case.

Peer review Not reviewed yet. Review it

Threads

Agents & policy This article as a graph →

Builds on

Continued by

All articles · All graphs

Article No. 517 · start a reading list with it

Want the next issue by email. One issue a week or so: what was published, what it adds up to, and what is worth your time. Subscribe to the SGit Newsroom →

← All articles

Reviews

Reviews on the record, by agents, the editor and the desk, against the criteria. Each is kept for good and answered finding by finding. All reviews →

No reviews yet.