Agents
Prospects increasingly ask an AI assistant before they ever visit a website. Spectare gives the assistant something better than a scraped paragraph: a page assembled for the person asking.
Spectare reads intent from referrer, UTM signals, and page context, then builds a unique content experience in real time, no code, no variants to maintain.
From: What is Spectare, Every visitor gets a different experience
The same content brain that personalises your website answers assistants. Nothing is written twice.
A user tells their assistant what they are looking for. The assistant passes that description to Spectare: who is asking, and what they want to know.
The same classification and assembly that runs on your website runs for the assistant: intent, audience, stage, and the atoms that answer them.
Assistants that support interactive apps render the page as a card in the conversation. Others share a personalised link. Either way, the answer is yours.
Spectare speaks the Model Context Protocol, the open standard AI assistants use to call tools. Any MCP client can connect to your workspace, ask about your product, and request a page assembled for a described visitor. No API key is needed to connect.
Your content answers inside ChatGPT, Claude, and every other assistant that speaks the protocol.
When an assistant that supports interactive apps asks for a page, Spectare returns the full assembled page and it renders as a card in the chat itself: headline, content, comparisons, stats, and working links back to your site.
The prospect sees a page built for them without leaving the conversation they asked in.
Spectare generates a curated, always-current map of your content at a standard address assistants already check. An AI reading your site gets your facts from a few structured fetches instead of scraping every page.
Assistants quote your content accurately, because you handed them the accurate version.
Mint a link for a person you already know something about: their role, their stage, even a link lifetime of up to 90 days for emails that sit in an inbox. When your tooling already knows the audience, it can state it outright rather than leaving it to inference.
The first click of a cold email lands on a page built for that reader, not your homepage.
Spectare's own workspace is live as an MCP connector. Add it to ChatGPT in developer mode and ask it to show you a page:
https://spectare.ai/api/mcp, no authenticationEvery workspace gets the same surface for its own content, on every plan: your connector URL is https://spectare.ai/api/mcp/{your-workspace-slug}. The API reference documents the tools an assistant can call.
Assistants get exactly the content humans get: the same published atoms, the same measured claims, arranged for whoever is asking. There is no separate script for robots, and nothing is shown to an assistant that a visitor could not read on your site.
That matters because assistants increasingly decide what to quote. The version of your product they describe should be the version you actually published.