VELORYZ STUDIO / KNOWLEDGE

Generate a site from a prompt

Brief, references, structure, generation and review.

Generate a site from a prompt

VELORYZ Studio · AI creation and prompt workflows Audience: Project brief authors, Studio operators and AI integration developers Guide type: Practical Studio operations guide

Purpose

Generate a site from a prompt should be treated as an end-to-end responsibility, not an isolated button. Start by defining what will change, who can approve it and how success will be checked. Brief, references, structure, generation and review.

The public creator helps produce a brief. Whether an autonomous site generation workflow exists must be verified separately; never describe a suggested design as already built.

Availability and responsibility

The source contains a project AI endpoint and protected AI connection settings; this does not prove an autonomous, self-service drag-and-drop AI editor or unlimited AI image generation. Provider credentials, quotas and model permissions must be configured and tested independently.

Before committing time or budget, confirm who operates this workflow: the visitor, the client portal user, a Studio owner, or the delivery team. A feature described in design documentation is not automatically a public account entitlement.

Before you begin

  • A clear goal, audience, constraints and reference content with permission to use it
  • A human review owner and acceptance criteria
  • Approved data-handling terms for material sent to an AI provider
  • A short list of measurable success criteria specifically for generate a site from a prompt.
  • A recovery plan if the proposed change creates a regression.

Recommended procedure

  • Locate the relevant area or public route. For new projects, establish the brief at /create/; for owner-only content changes, use authenticated Studio Control at /admin. Do not assume a missing feature exists.
  • Define the specific work item for Generate a site from a prompt: capture the initial state, user intent, relevant data, permissions, expected output and what must not change.
  • Follow the design and operating decisions listed below. Work in a draft, staging or reviewed project handoff where available; avoid editing live customer data experimentally.
  • Review the effect with real content and realistic viewport sizes. Record any exception and obtain the appropriate approval before a public change.
  • After the change, verify the public outcome rather than relying on a success message alone; preserve a reference to the reviewed version.

Topic-specific operating procedure

The following steps are specific to Generate a site from a prompt and supplement the general workflow above. Apply only the controls and roles actually available to your account or deployment.

  • Start with a detailed structured prompt: audience, pages, goals, style, content and technical constraints.
  • Treat generated sections or copy as a proposal to review; do not imply a ready-to-publish full site.
  • Check brand, copyright, privacy, feasibility and links before approving output.
  • Use the Studio delivery process to implement and release actual code.

Evidence to retain

  • A short before/after record of the actual user-visible change or reviewed deliverable.
  • The exact site URL, owner or reviewer, and the approval or release reference.
  • Any unresolved limitation, assigned follow-up, and a safe mitigation.

Configuration and decision points

  • Use /create/ to describe the project and review proposed scope before any implementation commitment.
  • Structure prompts around purpose, audience, pages, tone, constraints and measurable outcomes.
  • Treat generated suggestions as drafts and verify facts, links, rights and technical feasibility.
  • Administrators can test the actual AI provider connection from protected Studio settings.

When more than one implementation is possible, prefer the simplest approach that remains accessible, maintainable and compatible with the current source. Note any cost, licensing, privacy, ownership and performance implications before approval.

Verification checklist

  • The AI response never silently promises completion, price or publishing.
  • Generated content is accurate, original enough for intended use and reviewed for copyright concerns.
  • No private token, client secret or personal record is included in prompts without authorization.
  • Failures have a transparent explanation and a non-AI alternative.

Also confirm that the result addresses the original objective: Brief, references, structure, generation and review.. Review the customer-visible language, error states, navigation paths and the relevant device widths. A green build or a successful save does not by itself establish a correct production outcome.

Troubleshooting and recovery

  • If generation fails, retain the brief and retry with a smaller clear request.
  • If a configured provider rejects a model, check access and limits in protected settings.
  • If output contains invented facts, replace them with verified information before publication.

When asking for assistance, include the exact public URL or project reference, a concise reproduction path, the expected behaviour, actual behaviour, relevant timestamps and a redacted screenshot if useful. Never include passwords, access tokens or private customer records.

Related reading

  • [AI Studio overview](/docs/studio/ai/overview/)
  • [Generate or redesign a section](/docs/studio/ai/generate-section/)
  • [Edit selected elements with AI](/docs/studio/ai/edit-selection/)
  • [All VELORYZ documentation](/docs/)
  • [Studio documentation](/docs/studio/)
  • [Help Centre](/help/studio/)
  • [/docs/project-ai/](/docs/project-ai/)
  • [/create/](/create/)

Operational note

This article describes the Studio workflow and its acceptance criteria. Where a self-service control is not available, the action is coordinated with the Studio team rather than silently represented as automated.

Let’s talk
Make your project