Secure Agent Actions
Let AI assistants use your website safely, and prove it
This opens your contact forms to AI assistants, so they can send you enquiries on a customer’s behalf. Then we try to break it, and hand you a report showing exactly what is protected.
£1,500 fixed. Around five to seven working days.
Why this one needs proof
Anyone can switch it on. The value is proving it is safe.
The plumbing is a plugin: it can expose your forms and tools to AI assistants in a click. Left like that, it is on by default and untested, and you have no idea whether an anonymous request is refused, whether the limit that stops abuse actually works, or whether a private page stays private.
So the thing you receive is not a sentence like “your form is now callable by AI.” It is a proof report: a plain-English record of what we tried and what happened, including the things that were meant to fail.
What we prove
We do not just show it working. We show what happens when it should not.
It works
The tools are found, an assistant can submit your form, and the enquiry lands in the inbox you actually read, with every field intact.
It refuses
We call your enquiry tool with no credentials and record the refusal. A made-up token is rejected. Anything switched off stays off.
It holds up
The limit that stops abuse actually engages at the set number and recovers afterwards. Spam checks and honeypots fire as they should.
It stays private
Drafts, private pages, customer data and hidden fields are asked for through the tools, and not returned. We read a full response by hand to be sure.
What you receive
A report you can trust, not a promise
- The Proof Report. Two or three pages: every tool in plain English, what we tried to break, what stays private, and anything left open. It leads with the tests that were meant to fail.
- The health check. The Agent Simulator report as your technical appendix.
- A walkthrough call. We talk it through, and if you like, you watch an assistant use your site live.
- The agreement. A short document, signed before anything was switched on, setting out what an agent is allowed to do.
- A 30-day review. This matters most on this tier, because the first month is when real agent traffic actually turns up.
Agreed first, in writing
Nothing goes live until you have said yes to exactly what an agent can do
This tier makes software able to do things on your site, not just read it, so it deserves a conversation before, not an explanation after. Before we enable anything, we agree in writing what an assistant may do, what must stay private, where enquiries land, and where your off switch is.
An enquiry from an assistant is a request, exactly like one typed by a person. Nothing is booked, confirmed or charged, and nothing is final without a human step.
The honest part
What this does, and what it does not
We make your site safe to be used by an AI agent. We cannot control how any individual agent behaves. Testing proves your site behaved correctly on the day, under the conditions tested; it is not a security certificate and we will not call it one.
The request limit is shared across all your visitors, not per agent. And a plugin update, a theme change or a new caching layer can all quietly alter this, which is exactly what an ongoing re-test is for.
Where it fits
This comes after the groundwork, never instead of it
Your site has to be readable before actions on it are worth opening, so Secure Agent Actions is bought after or alongside the AI Visibility Sprint or the Foundation, never on its own.
Not sure where you stand? Start with the free scanner, or take the whole thing in one piece with the Sprint.
Pricing
Fixed price, no surprises
Secure Agent Actions
£1,500
Your lead capture, opened to AI and proven safe.
- Forms and tools configured to your agreed boundaries
- The full test plan run: it works, it refuses, it holds up, it stays private
- The Proof Report and Simulator health check
- The agreement, signed before switch-on
- A walkthrough call and a 30-day review
- Around five to seven working days
Ongoing quarterly re-testing is available on a retainer, because plugin and theme updates can change what is exposed.
Secure Agent Actions FAQs
The plumbing is a plugin, yes. Left alone it is on by default and untested. What you pay for is the setup to your boundaries, the deliberate break-testing, and a report that proves what is actually protected. The honest answer to “could I do it myself?” is: you could, and you would not know which of the safety checks your site currently fails.
No, unless you ask for it. By default it lets assistants send you enquiries, which are requests, not bookings or charges. Cart and checkout tools on a shop are a separate, explicit conversation we would rather have on a call.
An enquiry from an assistant is treated exactly like one typed by a person: a request you confirm before anything is agreed. Nothing an agent does is final without a human step.
Yes, and we show you where the off switch is before we finish, so you are never stuck with it. At least one person on your side can switch it off without us.
This comes after them, not instead. The Sprint and Foundation make your site readable and quotable. Secure Agent Actions opens your forms to assistants so they can act, and proves it is safe to do so.
Request your quote
Tell me a little about your website and I will send a precise, itemised quote, with no hidden steps and no obligation. If your site is not yet AI-ready, I will say so and point you to the right first step.