The pattern is already on serious landing pages
Product teams have started treating “ask AI about us” as a first-class CTA. Nucleus puts multi-assistant links on mynucleus.com/health. Framer deep-links ChatGPT and Perplexity with a long product-workflow prompt. Levels.fyi puts AI summary links in the footer. taap.it asks visitors to let ChatGPT, Claude, or Perplexity decide if the product fits.
These are not chatbot embeds. They are outbound handoffs — the same job ContinueWith is built for. The difference is that the industry default is still static HTML.
If your competitors already ship Ask ChatGPT buttons, the question is no longer whether to hand off — it is whether you will measure it.
The missed opportunity: static UX, zero AI-journey data
A static button answers one GTM fear: “buyers trust ChatGPT more than our landing copy.” It does nothing for the next questions: Which pages start AI research? Which assistants win? Which prompt converts curiosity into a useful session? What did the visitor actually want to ask?
Hardcoded prompts also freeze strategy. Changing the question means a design sprint. There is no {{url}} / {{title}} adaptation across pricing vs docs. There is no composer for the real objection — salary band, genetic risk, vs Linktree — that only the visitor knows.
From an analytics view, the best case is an outbound click event. That is not AI Journey Analytics. You cannot prioritize GEO or content work from a footer link click that looks like any other exit.
What “dynamic” means without hosting chat
Dynamic here does not mean another on-site chatbot. It means infrastructure at the handoff: dashboard prompt templates, optional free-text composer with LLM logos as send, page context variables, and continuation events at the website→AI boundary.
ContinueWith keeps the privacy model teams already chose when they shipped static links — conversations stay in ChatGPT or Claude. You gain the ops and data layer those links omit. Composer (opt-in) is how free-text intent enters the funnel without proxying tokens.
Upgrade path if you already have static buttons
Keep the placement and the narrative (“see what AI says”). Replace hardcoded anchors with the ContinueWith snippet. Port your best static prompt into a dashboard template; add one or two page-specific variants.
Turn on composer if visitors ask divergent questions on the same URL. Watch continuation events for a week: page, prompt label or question, assistant. Retire the prompt that nobody picks. Double down where handoffs cluster.
Full side-by-side: /compare/static-ai-buttons. Related: /compare/chatgpt-link for single-provider raw links, /learn/ai-journey-analytics for what to measure next.
FAQ
Common questions
- What is a static Ask-AI button?
- A hardcoded link that opens ChatGPT, Claude, or Perplexity with a prefilled ?q= (or ?prompt=) string. The UI can look polished — logos, “Ask ChatGPT” labels — but the prompt and assistants are fixed in HTML.
- What do teams miss by staying static?
- Visitor free-text questions, per-page prompt libraries, dashboard iteration without redesign publishes, assistant mix analytics, and a real AI-journey funnel (page → prompt → assistant). You validated trust in AI, then discarded the signal.
- How does ContinueWith differ without becoming a chatbot?
- ContinueWith stays a handoff layer: window.open into the visitor’s assistant. Composer mode adds optional free text; the dashboard owns wrappers and suggestions; events capture the transition — not hosted inference or transcript storage.
Next step
See it on your site
ContinueWith measures AI continuations at the website boundary — which assistant, which page, which prompt — without seeing private conversations. Install takes minutes.