-2.png?width=132&height=32&name=slatecx-logo%20(1)-2.png){kind=small}
Your content, turned into answers a buyer can actually use.
In one sentence: SlateCX turns your marketing content into guided, AI-assisted buyer journeys called Slates, then shows you what buyers actually engaged with. This page explains the mechanism end to end, including what the agent does when it does not know the answer.
The short version, all of it. You import your existing content (site pages, documents, HubSpot files); SlateCX turns it into reviewable cards and you assemble them into a campaign once. Buyers arrive through a widget, link, email, or QR code into a Slate: a persistent workspace where an agent answers their evaluation questions from your content only, shows the right card at the right moment, says so plainly when it does not know, and lets them save and return. Day to day, your marketer spends about fifteen minutes a week reading funnel and wave reports and adjusting content based on what buyers actually asked; your rep gets an alert with the full conversation context when a buyer signals readiness and starts the first call already briefed. What you stop doing: one-off landing pages, gated PDFs, and discovery calls that re-ask what the buyer already answered.
What your prospect experiences
A prospect meets SlateCX through an entry point: a widget on your site, a shared link, an email invitation, or a QR code. What opens is not a landing page and not a chat bubble. It is a Slate: a conversational workspace where they can ask real evaluation questions and get grounded answers, alongside the documents, pages, and next steps that answer them.
The buyer asks in natural language. The agent answers from your content, not from the open internet.
Relevant assets appear inside the conversation: PDFs, pages, videos, case studies, forms, meeting links.
Proactive prompts suggest a useful first question, so nobody stares at a blank box.
Buyers can save their Slate and return later. The conversation and the cards are still there.
How your content becomes the agent’s knowledge
Everything the agent says traces back to material you put in. There is no step where it improvises from general internet knowledge about your company.
Import sources
You bring in website pages, crawls, documents, files, and HubSpot files. SlateCX scans what you give it, identifies the useful items, and helps your team decide what should become campaign-ready content.
Create cards
A card is the unit of content a buyer sees: a brochure, a web page, a demo asset, a meeting link. Generated copy is written to read customer-facing, and your team reviews before anything goes live.
Assemble campaigns
Cards are organized into campaigns: the packaged experience for a specific audience, message, or goal. Each campaign carries its own knowledge and card library, and that is the boundary the agent answers within.
How the agent decides what to say and show
When a buyer asks a question, the agent answers from the campaign’s configured knowledge and surfaces the most relevant card. Cards can appear three ways, because not every buyer asks for the exact thing you want them to see:
A card is shown when the visitor asks about a known topic you have mapped to it.
The AI judges a card relevant to where the conversation is and surfaces it on its own.
A card can appear after a threshold of engagement, for example a meeting card after several substantive exchanges instead of waiting to be asked.
What happens when it does not know. The system is built to detect when no matching knowledge exists for a question, and to say so rather than improvise. It is also tuned to keep internal-only material out of buyer conversations, and it runs on fallback model chains, so if one AI provider has trouble, another takes over without the buyer noticing. An agent that admits the edges of its knowledge is the difference between an evaluation tool a buyer can trust and a chatbot performing confidence.
Waves: the opening move
A wave is a proactive prompt that offers a useful first question, like “Show me customer case studies” or “How does this work for financial services?” Clicking one is treated exactly as if the buyer typed it.
Dynamic Waves go one step further for anonymous visitors: using company-level identification signals, SlateCX can infer a broad business segment and generate a relevant opening question for that kind of company. The design rule is relevance without creepiness: the wave is tailored to a company type, never to a named person. The system does not greet visitors by name or reveal that it knows who they are.
How buyers enter, and how they come back
| Entry point | How it works | Typical use |
|---|---|---|
| Embedded widget | A Slate experience opens inside your website or a specific page. | Website conversion, product-page engagement |
| Shared link | A direct link opens a campaign experience immediately. | Sales follow-up, nurture, account sharing |
| Email invitation | An invited prospect receives a direct link into the right Slate, already knowing who they are. | Outbound follow-up, tailored outreach |
| QR code | Scanning opens a campaign-specific Slate. | Events, conferences, booth experiences |
Anonymous visitors are prompted to save their Slate and return later, which is how a first-visit conversation becomes a persistent evaluation workspace instead of a session that evaporates when the tab closes.
How HubSpot fits in
SlateCX is built for the HubSpot ecosystem and listed on the HubSpot Marketplace. In practice that means the Slate works alongside your CRM rather than beside it:
Public HubSpot files can be sent into a Slate as cards, so your existing asset library is the supply line.
Invitation and campaign flows connect known prospects into the right Slate; HubSpot activity and profile context can enrich follow-up.
Meeting and form actions are tracked and scoped so only intended actions count as conversion events. Engagement is measured honestly, not inflated.
What your team actually does with it
The honest answer is that most of the work is front-loaded, and the daily change is less about doing new things and more about acting on signals you never had.
Once, at setup
A marketer imports the sources (site pages, documents, HubSpot files), reviews what SlateCX turned into cards, edits the ones worth keeping, and assembles them into a campaign with its entry points and opening waves. This is content curation work, not technical work, and it is the bulk of the effort.
The weekly rhythm
Fifteen minutes in the funnel and wave reports: which Slates were created, which became engaged, where buyers drop, which opening prompts get clicked versus dismissed. Then small edits based on what you saw: sharpen a wave, add a card for a question buyers keep asking, retire one nobody opens. The loop is read evidence, adjust content.
When a buyer engages
This is the part that replaces work rather than adding it. When a buyer in a Slate signals readiness, your rep gets the alert with the whole context: what was asked, what was opened, where they hesitated. The first call starts briefed instead of opening with discovery questions the Slate already answered.
What you stop doing: building one-off landing pages per campaign, gating PDFs behind forms and hoping, and briefing reps from a form fill and a page-view trail. The content you already have does the qualifying; your calendar gets the buyers who are actually evaluating.
What you can measure
A Slate Funnel view shows where Slates sit in the journey, from initial touch to engagement to later outcomes: how many were created, how many became engaged, and where people drop. Wave Insights show which opening prompts get shown, clicked, ignored, or dismissed, and whether first messages came from a wave or from typing. The point is that you improve the journey from evidence about real buyer behavior, not from guesses.
Rollout discipline, stated plainly: new capabilities ship behind tenant-level feature flags and roll out gradually. Your campaigns do not change under you because we released something on a Tuesday. We would rather introduce functionality carefully than surprise a live buyer conversation.
Asked often, answered plainly
Is SlateCX just a chatbot?
No. Chat is one part of the experience. SlateCX also manages content cards, campaign entry points, proactive waves, follow-up flows, and reporting. The category difference is the workspace: a chatbot is a session that ends; a Slate is an evaluation the buyer keeps, returns to, and shares.
Do we need to configure AI models ourselves?
No. Model management is handled at the platform level, including fallback chains across providers. Your team works with content, campaigns, and prompts, not model settings.
Does it work for anonymous visitors or only known prospects?
Both. Known prospects enter through links, invitations, or CRM-connected flows. Anonymous visitors engage through embedded experiences and Dynamic Waves, and can save their Slate to make the relationship persistent.
What does the agent do when it cannot answer?
It detects that no matching knowledge exists and says so rather than improvising, and internal-only material stays out of buyer conversations. If you want to see this behavior yourself, ask a live Slate something obscure and watch what happens.
See the mechanism grade your own site
The Buyer Answer Grader runs the questions buyers actually ask against your website and against AI, and shows exactly where answers exist, where they are only claims, and where a Buyer Agent closes the gap.
Grade your site, free → Or request early access