How to Use AI to Set Up Your Site with the Onboarding Wizard

Onboarding Is Not Setup. It Is Context Building.

Most onboarding flows collect basic info to personalize the UI. Pixelesq's onboarding is different: the information you provide becomes the context the AI references every time it generates content for your project. Weeks after onboarding, when you ask the AI to write a blog post or draft a landing page, it uses your onboarding inputs to shape the output.

This means cheap onboarding inputs produce cheap AI output. Generic description equals generic content. Specific description equals specific content. The 5 minutes you spend here is the highest-leverage time in your entire project.

The Two Starting Paths: Import vs Describe

Step 1 gives you a choice: Have a website (import from URL) or Start from scratch (describe in 240 characters).

Import: Richer Context, Less Work

Paste your existing site URL. Pixelesq's AI fetches your pages, extracts your brand voice, messaging, visual identity, and content patterns. This gives the AI dozens of examples of how you talk about your product, what you emphasize, and what makes you different.

Import is the default recommendation if you have any existing web presence, even a basic landing page. The analysis takes 30-60 seconds and the context the AI gains is hard to replicate manually.

Describe: Works When You Do Not Have a Site

If you do not have an existing site (new business, rebranding, just launching), describe your project in up to 240 characters.

The difference between cheap and good descriptions:

• Cheap: "Marketing agency"
• Better: "Digital marketing agency for e-commerce brands"
• Good: "Full-service digital marketing agency specializing in paid media for DTC e-commerce brands under $10M ARR"

Each step up in specificity produces noticeably better AI output for the rest of the project. Spend 2-3 minutes writing a specific description instead of 10 seconds writing a generic one.

Industry Selection: The Biggest Tone Lever

The industry you pick is the single biggest input to the AI's default tone. 25+ options cover most verticals:

• Healthcare triggers clinical, trust-building, compliance-aware language
• SaaS & Cloud Services triggers product-focused, conversion-oriented copy
• E-commerce triggers catalog-oriented, price-aware, shopping-journey content
• Education & E-learning triggers pedagogical, clear, step-by-step explanations
• Legal, Financial, Real Estate each have their own vocabulary and tone

Pick the closest match even if it is imperfect. "Technology & Software" is a reasonable fallback for businesses that do not fit cleanly elsewhere. Industry affects every AI content generation downstream: blog posts, landing page copy, SEO descriptions, chatbot responses.

Primary Goal and Visitor Intent: CTA Intelligence

Step 3 asks you to pick a primary goal (Lead Generation, Direct Sales, Brand Awareness, Education, Customer Support, Community) and the visitor intents your audience has (Research, Compare, Purchase, Support, Pricing, Contact Sales, Download, Join Community).

This shapes the AI's CTA recommendations and content hierarchy. A Lead Gen site with "Compare Options" intent gets "See how we compare" CTAs. A Community site with "Join Community" intent gets "Join 10,000+ members" messaging.

Pick the goals that match your actual strategy, not the ones that sound ambitious. If you are bootstrapping with no sales team, Lead Generation is probably the right primary goal even if Direct Sales sounds more impressive.

Theme Configuration: Trust Defaults

Step 4 lets you pick an accent color (8 presets or custom), body font, and heading font (from 49 Google Fonts). If you imported an existing site, the AI pre-suggests colors that match your current brand.

For a first project, accept the defaults or the AI suggestions unless you have strong brand guidelines. You can iterate on the theme later with a live preview. Getting stuck on font choice here delays the part of the project that actually matters: building pages.

Feature Showcase and Final Options

The final step mostly informs you about Pixelesq capabilities. Two functional options: include a contact page and include a blog. Check both if you know you will use them. Skipping means you add them manually later.

How to Tell If the AI Has Good Context

After onboarding, test the AI before building real content:

1. Create a test page.
2. Add a hero section and click Generate (Magic Drop) to let the AI write the content.
3. Review the output. Does it sound like your brand? Does it reference your industry correctly? Does it target your stated audience?

If the output is generic, your onboarding inputs were too vague. Go to Settings > Website and rewrite your project description and tagline to be more specific. Changes propagate to future AI generations.

If the output is on-target, you did onboarding well. Build with confidence.

Updating Context After Onboarding

The onboarding wizard runs once. Everything it configures can be updated later:

• Project description, tagline, industry, goals live in Settings > Website
• Theme colors and fonts live in Themes
• Knowledge base (additional context the AI references) lives in Context in the sidebar

The Context section is especially useful for adding information the AI would not know from your website alone: internal documents, positioning memos, brand guidelines, style guides. Upload these and the AI references them during generation.

Pro Tips

• Write your description like you are pitching to an investor. Specific, punchy, differentiated. "We help X do Y without Z" formulas work well.
• Import even if you plan to rebuild. The AI context from an old site is valuable even if you throw away the visual design. You can reset the theme while keeping the brand voice context.
• Use the Context section as an ongoing brain. Add new documents, memos, and updates there as your business evolves. The AI references them the same way it references your onboarding context.
• Test AI output after changing onboarding settings. Changes to industry or goals affect future generations but not past ones. Generate new content to see the effect.
• Do not leave onboarding fields empty just to finish faster. Every blank field is a missed opportunity to steer the AI. The 30 seconds you save equals hours of fighting generic AI output later.

Troubleshooting

AI analysis of my imported site hangs: Takes 30-60 seconds for most sites. Large sites (50+ pages) can take up to 2 minutes. If it exceeds 2 minutes, refresh and switch to "Start from scratch" with a rich description. The import feature is a convenience, not a requirement.

Imported brand colors are wrong: The AI extracts colors from your existing site but does not always get brand colors right (hero images can confuse the extraction). Override in the theme step with correct hex codes. This is a one-click fix.

AI-generated content still feels generic after onboarding: Your inputs were too vague. Go to Settings > Website and rewrite your description, tagline, and goals with specific details. Also add context in the Context section. Re-run a Magic Drop generation to see if the output improved.

Skipped onboarding entirely: Configure everything manually. Settings > Website for description, industry, goals. Themes for colors and fonts. The AI still works; it just has less guided context than a completed onboarding provides.

Cannot update industry after onboarding: Go to Settings > Website. The industry dropdown is there. Changes affect future AI generations, not past ones. If your industry changed significantly (pivot), update this field immediately.

FAQ

Does the AI onboarding wizard use my data to train AI models?

No. Your onboarding inputs are used only within your Pixelesq project to give the AI context for generating your content. They are not used to train the underlying AI models. Your description, goals, and imported site content stay private to your account.

If I pivot my business, how do I update the AI's understanding?

Go to Settings > Website and update the project description, tagline, industry, and goals to reflect the new direction. Also update the Context section with any new documents or memos explaining the pivot. The AI references the updated context for all future content generation. Past generations are unaffected; you would manually review and regenerate old AI-generated content to match the new direction.

Can I use the onboarding wizard for a project that is not a traditional business website?

Yes. The wizard works for personal sites, portfolios, community hubs, nonprofits, events, and many other project types. Pick the closest industry match ("Other" is available as a fallback) and adjust the goals to match. The AI adapts to non-commercial contexts, though the default industry options are most refined for business use cases.

How does the AI use my imported site content behind the scenes?

When you import an existing site, Pixelesq fetches your pages and extracts brand voice (tone, vocabulary), key messaging (what you emphasize), visual identity (colors, fonts), and content structure (what sections you use). This becomes part of the AI's context. When you later ask the AI to generate content, it references patterns from your existing site to match your style. The original site content is not copied wholesale; it is used as a style and context reference.

Can I import content from a platform like Notion, Google Docs, or a PDF?

The onboarding wizard imports from public website URLs only. For importing content from other sources, use the Context section in the sidebar. You can upload documents, paste text, and link knowledge base content there. The AI references the Context section the same way it references onboarding inputs.

What happens if I onboard multiple projects with similar descriptions?

Each project gets its own independent context. The AI does not share context across projects, even if they are in the same account with similar descriptions. If you manage multiple similar projects, re-enter context in each onboarding or replicate via the Context section after. For agencies managing many similar client projects, this is one of the tedious parts of Pixelesq; there is no bulk-context feature yet.