A production-grade Teacher→Student system: a server-side Teacher (Torch GRU + contextual MLP with dual policy/reward heads) trains on rich synthetic + real clickstream DNA, while a tiny, quantized Student (<200KB, ONNX/WASM) runs in the browser to guide users in real time—privacy-first and performance-obsessed.
Already 100+ e-commerce stores signed up
Personalization in e-commerce today is broken. Current tools rely on server-heavy infrastructure, endless data pipelines, and invasive tracking that clash with privacy regulation and erode user trust.
Hundreds of KBs of scripts slow sites and break performance
$50k+/year SaaS contracts or full growth teams required
Reliant on cookies, cross-site IDs, or massive data exports
Most personalization = static carousels, not true behavioral guidance
Legacy players like Dynamic Yield, Optimizely, and Insider focus on enterprise server-heavy stacks. Shopify App Store is crowded with generic upsell plugins that don't adapt. Even AI newcomers are cloud-first wrappers, not local-first decision engines.
Patterncurve flips personalization on its head: a <200KB, local-first engine that learns in the browser, predicts in real time, and guides users with subtle nudges — all without servers, cookies, or bloat.
<200KB runtime, loads instantly
Runs locally, no PII ever leaves the device
One line of code, works like Clerk or Stripe
Micro-interactions that boost engagement and sales
Create personalization logic with our developer toolkit
Ship your configuration with the lightweight browser engine
AI learns locally and personalizes experiences instantly
A Teacher→Student architecture that creates organic buying guidance with behavioral realism—optimized for uplift, not surveillance—and deployable in a single line of JS.
Torch GRU + contextual MLP with dual policy/reward heads. Trained with CE + uplift + calibration losses across millions of simulated sessions.
<200KB ONNX/WASM model runs locally in the browser. Predicts next-best actions and renders subtle nudges in real time.
Hyper-realistic journeys with diverse entry/exit paths, product categories, and 40+ nudge types (urgency, trust, social proof, offers, exit-recovery) to learn humane behaviors.
Condition-driven eligibility: low stock, dwell time, delivery ETA, multi-PDP browsing—precise moments for effective nudges.
Multi-vertical catalogs with balanced pricing and stock distribution, validated schemas, and robust coverage.
Visual + console inspector to explore journeys like an analytics tool: see decisions, eligibility, and lift.
Validation & Uplift
Track AUC/accuracy, uplift estimation, and segment weighting during training.
Teacher Hub + CDN
Compile, roll out, and monitor Students per segment with safe deployment controls.
Privacy-First
Local learning avoids surveillance—no cookies, no PII exfiltration, no heavy pipelines.
A complete toolkit for privacy-first personalization
Node-first developer toolkit to define flows and actions
Lightweight browser runtime that predicts and nudges in real time
Analytics dashboard for uplift insights
Coming SoonNo-code builder for personalized nudges
FutureSubtle nudges that guide users without being intrusive
Remind users of the product they came for if they drift away
Auto-reopen the product users last engaged with when they return
Subtle breathing animation on Buy buttons to gently draw attention
Inject original context into carousels as users explore
Show savings when a user browses pricier alternatives
Keep original product always at hand with floating thumbnail
Three simple steps to privacy-first personalization
Use our Node.js toolkit to define personalization flows and actions
Deploy your configuration with Strand Client in the browser
{'<'}200KB AI learns and personalizes experiences instantly
Heavy scripts slow websites
Cookie-based personalization breaks trust
Expensive SaaS tools deliver little value
Third-party cookies are dying, ecommerce needs privacy-first personalization
<200KB runtime, DNA-based local learning, unique edge model
Works in Shopify to enterprise marketplaces, one line of code
+20% conversion uplift
Proven results
-50% infra cost
Lightweight solution
10x faster implementation
vs SaaS alternatives
npm install @patterncurve/strandimport { createHelix } from "@patterncurve/strand";
const hx = createHelix({ appId: "myshop" });hx.track("view:pdp", { page: "pdp" });const next = hx.predict({ k: 2 });
console.log(next);
// => [{ id: "cta:add_to_cart", score: 0.87 }]hx.on("suggest:cta:add_to_cart", () => {
hx.highlight("#buyBtn");
});npm install @patterncurve/strandDefine flows once, run anywhere
Hook into loyalty APIs, offers, badges
Works even without network
Learns locally in real time
Create personalization logic
Ship to production
Real-time personalization
Be among the first to experience privacy-first personalization that actually works.
Join 100+ e-commerce stores already signed up
Let's discuss the future of personalization.