Buyers increasingly research with AI, and agents increasingly do the visiting. Spectare turns your messaging into one structured content library that answers both: a page assembled in real time for each person, and clean, current, machine-readable content for the agents doing the research.
No segments, no stored profiles, no training period. Works from visitor one.
Spectare reads visitor intent and assembles personalised content in real time, without code changes or developer involvement.
Content assembled for you based on how you arrived.
Built on Claude
Classification, brand voice, assembly, experimentation, and quality review: all automated.
Spectare reads referrer, UTM params, and page context on every visit and classifies intent with Claude in real time. No training period, no historical data. Accurate from visitor one.
Write once how Claude should sound: direct, formal, conversational, technical. Claude applies your instructions when writing headlines and CTAs in hero and image components. Every visitor, same voice.
Set an exploration rate and Spectare tests atom arrangements and images on a share of visitors. Every conversion feeds the result back into ranking alongside semantic similarity, so assembly gets more precise with each visit.
Describe the atom you need. Claude generates the title, content, audience, stage, and tags. You review and save. The bottleneck moves from writing to editing.
Natural language summaries of conversion and assembly performance, surfaced in the analytics dashboard. See what is working without building a report.
When you submit an atom, Spectare checks it for style violations, missing stats, and metadata gaps automatically. Issues surface inline on the atom list without opening the editor.
Three lines of code
No SDK to configure. No segments to define. No training data to collect. Drop in the script, mark your content slots, and Spectare handles the rest. Your existing design, personalised from the first visitor.
<!-- 1. Drop in the script -->
<script
src="https://spectare.ai/spectare.js"
data-org="your-org-slug"
></script>
<!-- 2. Add slots wherever you want personalised content -->
<div data-spectare-slot="primary"></div>
<div data-spectare-slot="supporting"></div>
<!-- 3. Set your conversion URL in Settings. Fires automatically. -->
<!-- Or call spectare('conversion') manually on any page -->Segmentation
Sorts visitors into a handful of predefined buckets. Everyone in a bucket gets the same experience, decided in advance.
Assembly
Reads this visitor’s signals in the moment and assembles their page from your content library. Every visit is its own decision.
No configuration. No segments. No training data. Works from visitor one, learns from every visit after.
One script tag (or the WordPress plugin). Spectare reads referrer, UTM params, and page context silently, with no forms and no questions asked of your visitors.
Spectare classifies the visitor's intent in real time using Claude. No training period, no historical data required. Works on day one.
New intent patterns stream progressively into your slots in under 200ms. Recognised patterns arrive inlined in the HTML itself, with zero API calls and no flash of generic content.
Every conversion tells Spectare what worked. Assemblies that convert rise. Arrangements are tested automatically at your chosen exploration rate. The more visits, the more precisely Spectare selects.
Analytics
Set a conversion URL or call one line of code. Spectare records which atoms and image every converting visitor saw, builds a co-conversion graph of what performs together, and feeds it straight back into ranking.
Pro plans get natural language summaries of performance and content gap detection in the dashboard, no reporting setup required.
See analytics →Co-conversion graph
87 co-conversions
74 co-conversions
Rebuilt nightly from 90 days of conversion data
Every visitor gets a personalised experience built from the same source material. When someone arrives, Spectare classifies their intent from referrer URLs, UTM parameters, and session context, then passes a context token describing their audience, buying stage, and confidence level to the assembly step.
Assembly runs on the server. Claude receives the context token alongside three ranked inputs: a vector search over your atom library finding the most semantically relevant candidates, a conversion history showing which atoms and arrangements have converted for this audience and stage before, and an image selection picking the best visual for the moment. From these signals Claude selects which atoms to include and maps them to one of nine component types.
Two of those nine components are written entirely by Claude in your brand voice: HeroStatement and ImageCtaHero. The remaining seven components, including StatGrid, FeatureList, TestimonialCard, ComparisonTable, CodeBlock, AtomCard, and CtaStrip, are drawn directly from your atoms without modification. Your words reach the visitor unchanged; Claude only writes the headline and hero layer.
The assembled page is cached at the edge so subsequent visitors with the same intent pattern are served instantly without another AI call. Every impression is recorded against the variant, audience, and stage that produced it, feeding back into the conversion history that informs the next assembly decision.
Spectare brings real conversion history into every page assembly decision. When the AI selects which content atoms to show a visitor, it weighs which atoms have actually converted for visitors with the same audience type and buying stage in past sessions. Content that has a proven track record of driving action is ranked higher for similar future visitors.
Every conversion event you record tells Spectare which atoms and images were shown during that session. This data feeds directly into the assembly algorithm, so the system gets more accurate the more traffic your site receives. Early on, Spectare relies more heavily on semantic similarity to match content. Over time, conversion history becomes the dominant signal for well-trafficked audience and stage combinations.
Tracking a conversion requires either setting a thank-you URL pattern in your dashboard settings, which fires automatically when a visitor reaches that page, or calling the conversion method manually from your own code for more complex flows such as multi-step forms or in-app upgrades. The conversion guide describes both paths.
The result is a personalisation system that continuously improves without any manual tuning. You write the atoms. Spectare learns which ones work for which visitors and adjusts every future assembly accordingly.
Everything you need to personalise at scale, without the complexity.
Script tag, WordPress plugin, or Next.js SSR hook. Drop it into WordPress, Webflow, Shopify, or anything that runs JavaScript.
Set a URL or call one line of code. Every conversion is recorded against the atoms and image the visitor saw, no separate analytics setup required.
No training period, no historical data. Spectare uses Claude to classify intent in real time, so personalisation works on day one, not after weeks of traffic collection.
No cookies, no visitor profiles. Intent is classified fresh each session from page signals, never stored against a person.
Write focused, reusable content blocks. Update once and every assembled experience stays current, no republishing required.
Recognised intent patterns are inlined into the HTML before JavaScript runs, with zero API calls and no flash. New patterns stream progressively, first section in under 150ms.
14-day trial, no credit card. Upgrade when you're ready.
Trial
$0 for 14 days
Starter
$19/month
Pro
$79/month
Enterprise
Custom
AI assemblies are cached. Once Spectare builds an experience for a visitor intent pattern, every matching visitor gets it instantly at no extra cost. Most orgs serve far more visitors than their monthly limit suggests.
Start your 14-day trial. No credit card required.
Create your workspace