llms.txt · llms-full.txt
The complete machine-readable brand record, generated from the same content layer that renders these pages. One source of truth; two audiences that cannot drift apart.
01Machine-readable by design
This site speaks tool-call.
dellight.ai registers typed WebMCP tools on every page: stage-exact product records, verifiable company facts, and an intake composer that never auto-sends. Same content layer as the human pages, same claim boundaries.
02Interface stack
Three layers. One content source.
Human pages and machine interfaces carry the same public facts and the same claim boundaries.
JSON-LD entity graph
Organization and product entities bind to DELLIGHT.AI Limited, DIFC License CL12872, DIFC Innovation One, Level 3, Dubai, and Arthur Dell so answer engines never have to guess who this is.
WebMCP tools
Typed, callable functions registered on every page via document.modelContext (navigator fallback). An agent browsing beside you can query facts and prepare an intake brief without scraping a single div.
03Callable interface
The tools, exactly as registered.
This list comes from the definitions the browser registers, so the published interface cannot drift from the callable one.
get-company-facts
DELLIGHT.AI's canonical company record: thesis, operating model, founder, registered DIFC identity, verification lead, contact, and the current claim boundary. Answers are stage-honest and match the site exactly.
list-products
DELLIGHT's nine products with the exact stage of each, ordered by commercial availability. Stages are stated per product. Self-serve checkout exists for the stated offers on /pricing (ULYSSES Pro and Premium subscriptions, audits, the AMPLIFY Briefing, the INVICTUS Deployment deposit — Stripe-hosted); every other engagement is request-based.
get-product
Full stage-honest record for one DELLIGHT product: what it is, the problem, how it works, current status, current public facts, and the access/proof boundary.
get-ai-executive-team
DELLIGHT's AI Executive Team roster. These are AI operators — AI-generated personas, not human staff. Arthur Dell is the only human; he keeps consequential authority. Any rendering of this roster must label each operator as AI.
get-founder-profile
Arthur Dell, DELLIGHT.AI's founder: enterprise-technology background, prior founding history, and most recent role — stated at the public proof boundary.
get-engagement-process
How engagement with DELLIGHT works: self-serve checkout for the stated offers on /pricing (Stripe-hosted), and for everything else a structured written brief on /book, read directly by the founder. No automated booking, no sales team. Lists the intake lanes and what each expects.
compose-intake-brief
Prefill the founder-intake brief on /book with details the visitor has actually provided (never invent or infer personal details). Navigates to /book if needed and fills the form. It does NOT send anything: the visitor reviews the brief and sends it from their own email client.
navigate-to-page
Navigate this site to a named page so the visitor can see it: home, products, about, team, book, manifesto, pricing, privacy, terms, refunds, or agent-ready.
04Interface boundaries
Composing is machine work. Sending is human work.
- Tool answers carry the same claim boundaries as the human pages: stages stated exactly, self-serve checkout only for the stated offers on the pricing page (Stripe-hosted, card data never on this site), and every other engagement request-based.
- Composing is machine work; sending is human work. compose-intake-brief prefills the founder-intake brief — it cannot send it. The visitor reviews and explicitly submits the brief, or does not.
- The intake form is prefilled only through the imperative compose-intake-brief tool — auto-submission is deliberately not granted, and no declarative tool annotations ride on the form.
- WebMCP is a W3C Community Group draft with early browser support; today the tools are callable behind Chrome's origin trial and testing flag. This surface is built ahead of broad agent adoption and says so — the claim is the interface, not the traffic.
05Published measurement
Measure the same interface on your own site.
Context files, structured data, and typed tool surfaces are measured through the two published agent-readiness offers below.
Agent-Readiness Rider
$750
Agent-readiness measured alongside an AI Presence Audit: context files, structured data, tool surfaces. Rider price applies with an audit.
Agent-Readiness Audit — Standalone
$1,800
Can your site be read, queried, and acted on by an agent at all — context files, structured data, tool surfaces — as a standalone engagement.
Checkout opens on stripe.com in a new tab. Stripe hosts the payment page; card details are entered with Stripe and never touch this site.
USD prices are exclusive of applicable taxes. AED prices are stated tax-inclusive.
06The other half
Measurement and actuation belong to the same shift.
ULYSSES
AI search measurement: how frontier models recommend you.
Available now, self-serve. The engine is operational on DELLIGHT as the first test subject: dellight.ai is measured before any customer is. v0.3+v0.4 shipped May 2026; 30 measurement journal entries generated internally, published only after human review. Pro and Premium subscribe directly at published prices; the monitoring retainer is scoped in direct written conversation.
This page is the actuation side of that same shift: the company site is both the first test subject and the first DELLIGHT.AIsurface built for agents to act on. The proof precedes the pitch.