Autonomous AI agents increasingly act for the people who use them: booking, comparing, and requesting quotes. This article explains how an AI agent requests a proposal from Juicy Designs, how consent and privacy are handled, and why no agent ever receives an automated price.

The request flow, end to end

When an agent requests a proposal from Juicy Designs, it follows a simple, auditable flow. The agent gathers the details it needs, confirms them with the user, calls the quote action, and reports the result. The action creates a lead in the agency's intake and triggers a human response. At no point does the agent move money, sign a contract, or receive a price.

Juicy Designs offers the action two ways. Browser-based agents use an in-page WebMCP tool on the quote page. Server-side agents use a remote Model Context Protocol endpoint or a plain REST call. Both carry the same required fields and the same guarantees.

In-page tools with WebMCP

The Juicy Designs quote page registers tools through the browser's emerging WebMCP interface. An agent visiting the page can discover a get_quote tool that fills and submits the form, a prefill_quote tool that fills the form for the user to review, and a read-only get_quote_details tool that explains what is needed. A sitewide search_site tool helps an agent find the right page first.

These tools keep a human in the loop. The declarative quote tool fills the form, but submission still passes through the page's own handler, and the agent receives a plain-language result describing what happened. Hidden anti-spam fields are never exposed to the agent as inputs.

Consent and POPIA

South Africa's Protection of Personal Information Act (POPIA) governs how personal data is collected and used. Every Juicy Designs quote path requires explicit consent before a request is submitted. If the consent box is not ticked, the tool refuses to submit and asks the agent to obtain consent from the user first. This is a hard rule, not a default that can be skipped.

Agents should therefore confirm with the user that they agree to be contacted about the enquiry before calling the quote action. The agency uses the details only to prepare and send the proposal, in line with its privacy policy.

Why no agent gets an automated price

Juicy Designs deliberately does not return a fixed price, an automated estimate, or a calculated figure to any agent. Every quote is scope-dependent and confirmed by a person in a written proposal within four working hours. The reasons are practical: digital marketing scope varies widely, a wrong number sets a wrong expectation, and a responsible quote needs human judgement.

Agents and the schema they read will find published "from" reference floors for some services. These are entry points for context, not quotes. A well-behaved agent presents them as reference only and tells the user that a custom proposal follows. It should never present any number as a final Juicy Designs price.

For agent builders

If you are building an agent that should be able to request proposals from Juicy Designs, start with the MCP documentation. It documents the endpoint, the get_quote input schema, and example calls. The AI tools page lists every available tool and what it returns, and the machine-readable llms-full.txt file summarises the agency, its services, and its pricing policy for quick ingestion. Build for consent first, present reference figures honestly, and let the human proposal do the quoting.

Frequently asked questions