A 5-minute tour of the auto-setup wizard

By Mike Berris · April 24, 2026 · 5 min read

The easiest way to understand Wengrow is to watch it go live on a real URL. This post is a text tour. If you'd rather see the video, it's embedded on the home page.

Let's use a made-up business — bayside-physio.com, a physical therapy clinic in Tampa. Pretend they have a reasonably content-rich website: a home page, a services page, a "meet the team" page, an FAQ, and a contact form. They've never used a chatbot before.

Minute 1: Signup

The prospect types bayside-physio.com into the Wengrow signup form as their business URL. That's the only required input. They pick a plan (let's say Growth, $499/mo). Enter a credit card for the 14-day trial.

Submit.

Minute 2: Crawling

Wengrow's crawler starts on bayside-physio.com. It respects robots.txt. It rate-limits so the small-business host doesn't see a surge of traffic. It reads the home page, discovers internal links, follows them up to a configured depth, and pulls down the HTML.

In real time, the signup screen shows what's being crawled. The prospect sees "Reading home page... Reading /services... Reading /team... Reading /faq..."

Total content ingested: about 40 pages of text, mostly service descriptions, pricing info, therapist bios, and FAQ entries.

Minute 3: Drafting the agent

While the crawler is still finishing, our AI starts reading what's been ingested. It identifies:

• The business category (physical therapy / healthcare practice).
• The service lines (back pain, knee injury, post-surgical rehab, sports therapy, dry needling, aquatic therapy).
• The tone of the existing website copy (warm, patient-focused, explicitly Tampa-based).
• The kinds of questions patients probably ask (insurance accepted, first-visit expectations, session pricing, location/parking).

Based on that, it drafts:

• Personality preset: Warm (appropriate for healthcare).
• Qualification fields: reason for visit (service category, not specific symptoms — healthcare limitation), insurance, zip code, preferred location, new-patient vs existing.
• Welcome message: "Hi! I'm Bayside's scheduling assistant. I can help you find the right therapist or book a first appointment. What brings you in today?"

All of this is a draft the prospect will tune. But it's a sensible, specific, industry-aware starting point.

Minute 4: Building the hosted page

In parallel with the agent draft, Wengrow assembles a hosted landing page at bayside-physio.wengrow.app. Sections:

• Hero: "Get back to moving without pain — Tampa's trusted physical therapy practice." (Drafted from the home page H1 + meta description.)
• Services: a grid of the 6 service lines, with 1-2 sentence descriptions pulled from the services page.
• About: a paragraph about the practice, with the therapist-team info condensed.
• Contact: a form + full-page chat widget.
• SEO metadata: title, description, Open Graph image auto-generated.

Branding colors are extracted from the existing website's CSS. Logo is pulled from the favicon and top-nav image. The prospect can swap any of this later.

Minute 5: Legal drafts + going live

Wengrow drafts a Privacy Policy and Terms of Service from a healthcare-practice template, customized with the practice name, business address, and state. Flagged clearly as "DRAFT — review with counsel before publishing."

The knowledge base finishes indexing. Vector embeddings are generated. BM25 index is built. Hybrid retrieval is live.

The prospect clicks "Finish setup."

bayside-physio.wengrow.app is now a live URL. Wengrow's bot is embedded on it. The widget is ready to embed on their actual website with one line of JavaScript.

Total elapsed time from "submit signup": 4 minutes 50 seconds.

What the prospect does next

This is the moment most chatbot platforms fail: the prospect is live, but what do they actually do first?

The admin panel opens on a guided tour:

1. Review your qualification fields — here's what we drafted. Tune them.
2. Test your agent — open your hosted page, ask a question, see how it responds.
3. Connect a CRM (optional) — Salesforce, HubSpot, or skip for now and use built-in lead management.
4. Embed on your real website — copy this script tag, paste before </body>.
5. Approve the legal drafts — or swap them for your own.

Most prospects spend the next 20-30 minutes doing steps 1-3, then schedule step 4 for when their web developer is available. They're using the product from minute 5.

What doesn't work

Let's be honest: the 5-minute setup doesn't get you 100% of the way to production-quality.

On well-documented websites (clear services, detailed FAQs, published pricing), you get to 80-90% of production quality in the first 5 minutes. Tune for 30 minutes, and you're production-ready.

On lightly-documented sites (a one-page brochure with a phone number), you get maybe 50-60% production quality. Tune for an hour or two, upload some supporting documents or Q&A pairs, and you're production-ready.

On sites with no real content at all ("coming soon" page), auto-setup isn't magic — the bot doesn't know anything to say. Those prospects use the hosted page + a manual description of their business to build up from scratch.

The wizard is the product

A lot of chatbot platforms have a wizard that collects settings and then dumps you into an empty flow editor. That's not a wizard — that's a setup form.

Wengrow's wizard actually builds the thing. You leave with a working product, not a blank canvas. That's the single most important UX decision we've made, and it's what the 5-minute claim is built on.