Turn Your
Website Into
an Expert
Support
Agent.Instantly.
Train WebsiteGPT on your site, docs, files, and manual Q&A. Launch a branded support widget, collect follow-up requests, and review every grounded answer from one operational surface.
Built for grounded support, lead capture, and review.
AI Support Bot
Online
Ask a question, collect context, or route to a human.
Cited
Answers tied to live sources
1 tag
To launch the widget
Q&A
Manual coverage for key gaps
Review
Feedback and unresolved queues
Trains on
Up and running in three steps.
The design follows the reference, but the product story stays anchored to what WebsiteGPT already ships today.
Sync your support source of truth
Start with a root URL, then layer in PDFs, markdown, and manual Q&A so the first launch already covers setup, pricing, and escalation paths.
Shape the assistant before it goes live
Adjust tone, allowed domains, blocked topics, greeting copy, launcher text, and fallback behavior in the same settings studio.
Embed, monitor, and improve
Ship a single widget script, collect leads in chat, route humans in when needed, and use conversation review to close knowledge gaps fast.
Everything your support rollout needs.
This keeps the reference page’s density, but replaces the placeholder SaaS feature list with real WebsiteGPT capabilities that exist across the widget, dashboard, and billing surface.
Grounded answers from real content
Website crawls, file uploads, and manual Q&A turn support content into answers that carry citations instead of vague summaries.
Lead capture inside the chat flow
When a visitor needs follow-up, the widget can collect contact details and attach the conversation so the next human starts with context.
Human escalation without chaos
Fallback behavior, escalation copy, and routed follow-up keep the assistant useful even when a question needs a person.
Safe rollout controls
Allowed domains and blocked topics define where the bot appears and which categories it should avoid before risk spills outward.
Quality and operator visibility
Helpful votes, unresolved conversations, top unanswered questions, and deflection signals help owners see where support still leaks.
API and workspace controls
API keys, billing state, branding controls, and team roles keep production rollout tied to clear limits instead of loose experiments.
Built for teams that need answers and oversight.
WebsiteGPT is not just the widget. The surrounding controls matter just as much: source readiness, escalation handling, quality review, plan enforcement, and rollout safety.
Source sync controls+
Owners can resync website and file sources instead of waiting for stale content to create broken answers.
Widget settings studio+
Launcher display, greeting copy, avatar, theme color, and fallback behavior are configurable before launch.
Quality queue+
Low-quality threads can be reviewed, saved into manual Q&A, or escalated when the support boundary needs a human.
Everything you need to roll out your own AI support agent.
The reference design uses a capability matrix here, so this version does too, but only with capabilities visible in the current codebase and product shells.
Connect the widget to the rest of your support workflow.
The reference shows integrations and a live demo. Here the page keeps that same rhythm with WebsiteGPT’s real operational surfaces: coverage planning, escalation design, and free tools.
Launch with the questions you cannot afford to miss
Seed manual Q&A for onboarding, pricing, and escalation before large crawls are fully tuned. The product already gives operators a place to patch those gaps.
See coverage guidanceMove from self-serve answers to human follow-up cleanly
The widget supports lead capture and escalation requests, while ops views preserve the thread, citations, and feedback around every conversation.
See escalation flowsModel the business case before rollout
Free planning tools help estimate deflection, support savings, source volume, and launch readiness before you commit to a larger deployment.
Open free toolsSee the production support experience before you ship it.
The settings studio already previews launcher, chat, and escalate states. This homepage mirrors that idea with a static support snapshot that keeps the reference layout language intact.
WebsiteGPT Demo
Online
Pricing
Pricing shaped for grounded support operations.
Keep the clean three-tier layout from the reference while tying the plans to actual WebsiteGPT rollout limits: coverage depth, operator visibility, branding control, and message capacity.
For single products
Starter
Launch one support assistant with guarded usage and tight follow-up loops.
For scaling support teams
Growth
Scale grounded support coverage across docs, product surfaces, and multiple operators.
For org-wide rollout
Scale
Higher-volume automation with stricter operational controls and more support coverage.
Support automation only works if the operating model holds.
Instead of fabricated testimonials, this section stays honest and uses the same three-card visual pattern to explain the real outcomes WebsiteGPT is optimized for.
Answer with evidence, not with confidence theater
The support assistant is built around retrieved content and visible citations so operators can verify what the bot actually used.
Keep the operator loop calm
When a conversation drifts beyond what the bot should handle, escalation and follow-up paths stay inside the same operational surface.
See quality regressions before customers do
Helpful feedback, unresolved rate, and repeated unanswered questions make it easier to tighten weak spots instead of guessing what broke.
Questions that usually come up before rollout.
The page keeps the reference’s accordion finish, but the copy points back to the current ingestion, widget, analytics, and tools capabilities in this repo.
Explore use casesWhat can I train WebsiteGPT on today?+
You can crawl a website, upload files like PDF, TXT, and Markdown, and add manual Q&A entries for the first high-value questions that need guaranteed coverage.
How does the bot stay grounded?+
Assistant replies are built from retrieved content and can render citations directly in the chat experience, which helps owners inspect whether an answer came from trusted source material.
Can I control where the widget appears?+
Yes. The chatbot settings studio exposes allowed domains so the widget can be constrained to the sites and surfaces you actually intend to support.
What happens when the bot cannot solve something?+
Fallback behavior and the lead capture flow let visitors request human follow-up. The conversation, supporting context, and feedback stay attached for the team.
Do I get visibility after launch?+
The dashboard already tracks unresolved rate, lead conversion, helpful feedback, top unanswered questions, source readiness, and broader support-health signals.
Can I test the economics before I roll this out widely?+
Yes. The free support ROI calculator and coverage planner are there specifically to estimate support savings, source volume, and launch readiness before scale-up.
Start building your grounded AI support layer today.
Crawl the site, seed the critical Q&A, install one script, and keep tightening quality from the same dashboard once real conversations start landing.