Consulting and AI SystemsOne doctrine, two linesEngagements are scoped and proposal-led
The corner of a severe unadorned ashlar building in flat daylight, its tall windows dark and its facade carrying no signage

AI Systems

Presence Systems

The buyer assembles a shortlist before contact, increasingly with a machine reading on their behalf. Presence Systems makes the business legible to both — the person deciding and the system summarising you to them.

The modules

  • AI-native site architectureA site built so both readers can use it: the person deciding, and the model summarising you to them. Semantic structure, the firm’s stated facts available in machine-readable form, and pages that answer one question completely instead of gesturing at an enquiry form.
  • Answer-engine visibility — schema, llms.txt, bot policyStructured data so an engine can state what you do without inferring it, an llms.txt that tells a model what is true and what is deliberately not claimed, and an explicit crawler policy deciding who may read and who may train. This site runs all three on itself, which is the only demonstration of the module we are prepared to point at.
  • Brand assets produced to a controlled identity systemIdentity work produced against a written standard rather than per-request taste, so a mark, a document and a page made months apart still read as the same firm. The standard is the deliverable; the assets are what it produces.
  • A diagnostic concierge on the site itselfA guided path that asks what the visitor is actually trying to establish and answers from stated material, rather than a chat window that improvises. It hands to a person at the point where judgment starts, which is the same rule the rest of the firm runs on.
  • Conversion engineeringThe measurable half: what each page asks for, where an enquiry goes, how it is qualified before it reaches a person, and what is recorded so the next revision is a decision rather than a preference.

Which of these is built, and to what depth, is decided by the diagnosis and stated in the proposal. None of it is sold as a fixed bundle.

Where this has been proven

An in-house beta, and we will say so.

Presence Systems is being built and run on Bastion’s own properties first — internal standard-operating-procedure development, not a client engagement. That is the honest status of the cluster today, and it is stated here for the same reason the record on the About page is the founder’s own and not the firm’s: our own source documents set the bar, and this cluster has not cleared the one that would let us call it a track record.

Status
In-house beta
Where
Running on Bastion’s own properties
What it is not
Not yet a client engagement

The threshold the cluster has to clear before it is sold on its own terms is written down internally: repeated instances at volume, a stable schema, and clients paying for it separately. Until then it is bought inside Foundry or Watch.

A dressed stone channel that ends abruptly at a squared kerb, the ground beyond it left rough

The rule

AI automates production, not decision-making or execution that requires judgment.

Inside this cluster that means the production work runs continuously and the decisions do not. A machine drafts, routes, qualifies and schedules. A person decides what is worth doing, and what to do when the situation is not the one the rule was written for.

Where it sits

Line
AI SystemsProduction capacity, governed.
Bought inside
Foundry or WatchNot licensed on its own.
Paired with
ConsultingThere is no point running an engine against a strategy that has not been diagnosed.

Ask what this would run for you.

The proposal states the modules, the scope and the fee, after the diagnosis.

Sent straight to the two people who answer it. No autoresponder sequence, no call centre.

Ask Bastion

Ask Bastion