01: READ
~4ms
Collect the session
The SDK gathers device, network, timing, geography, referrer, and behavioural properties already present in the request and the browser. No identifiers are created and no personal fields are collected.
Zero cookies.
Infinite personalization.
Most people who land on your store never sign in. Pulse reads their live session signals, works out what they came for, and personalizes the experience in about 50 milliseconds. No cookies. No logins. Nothing personal stored.
One Engine.
Every Industry.
The same real-time inference technology, adapted for any digital experience. Pulse reads behavioral signals to understand intent, no matter the platform.
The full personalization stack for online retail
Fully personalised home page
A few steps to
personalization.
Pulse builds two vectors — one for the shopper in front of you, one for every product you sell — and matches them. Neither side needs a history, which is why it works on the first page view.
1Send data
Real-time
Anonymous session
1,000+ signals · every request
Once at sync
Your catalogue
synced once, updated on change
Not required
CookiesLoginUser ProfileConsent gatePurchase HistoryTraining Window
2Build the twin
Visitor persona
Bayesian scoring
readiness · price band · category affinity · temp
→ 384–dim vector
Product twin
Persona-fit enrichment
suits[] · persona_fit{} · one twin per SKU
→ 384–dim vector
Vector match
cosine similarity ranked by fit score
<50ms
p95 · one round trip
0 cookies · 0 PII
3Get personalization
- Personalised homepage
- Search & category ranking
- Product Recommendation
- Cart & Checkout upsell
- Similar Products
- Email & retention
- Paid landing pages
- Mobile app feed
- ····Anywhere you render products
Under 50ms,
start to finish.
Four stages run on every request. Read the session, score the intent, rank your catalogue, return the payload — all inside one round trip, before your page paints.
02: INFER
~22ms
Score intent
The inference engine maps that signal set onto a live intent profile: category affinity, price sensitivity, purchase readiness, and session temperature. Cold-start included, because there was never a warm start to wait for.
03: RANK
~11ms
Order your catalogue
Your products are scored against that profile and returned in rank order with a match score on each. You keep full control of merchandising rules, inventory constraints, and margin logic.
04: RETURN
<50ms total
Render personalized
One JSON payload, fast enough to use before first paint. Render it in your grid, your search results, your recommendation slots, or your email — the response is yours to place.
One call. Ten lines.
Same afternoon.
REST and SDKs for JavaScript, Python, PHP, and Go. If you'd rather not touch code, our Shopify and WooCommerce apps do the same thing with an install button.
import Pulse from "@pulse-ai/sdk"; const pulse = new Pulse(process.env.PULSE_KEY); // no cookie, no login — just the live session const { persona, recommendations } = await pulse.recommend({ session_id: session.id, type: "for_you", limit: 12 }); render(recommendations.product_ids);
Response · 200 OK
41ms
{
"recommendations": {
"product_ids": ["prod_789", "prod_012", "prod_344"],
"rerank_scores": [0.94, 0.91, 0.77]
},
"persona": {
"primary": "value_conscious_runner",
"confidence": 0.89,
"price_band": "mid_upper",
"readiness": "comparing"
},
"metadata": {
"inference_time_ms": 28,
"signals_read": 1043,
"cookies_set": 0
}
}What we expect it
to move.
Figures below come from a simulated A/B test across 5,000 sessions over seven days. They are projections from our own model, not merchant results — we'd rather show you the method than borrow someone else's numbers.
+29%
Product CTR
+29%
Add to cart
+33%
Conversion rate
+33%
Revenue / visitor
Trusted by
Growth-Focused Teams
Commerce, streaming, and media teams all hit the same wall — personalization that only works for shoppers you can already identify. These are the results after switching to inference on the live session, no profiles required.
+38% Conversion
“Pulse understood that a visitor from TikTok at midnight has completely different intent than someone from Google at 9 AM. Our conversion lift was immediate.”
Head of E-Commerce
Fashion D2C Brand
+40% Session Time
“We plugged Pulse into our feed ranking and saw session times jump within a week. The privacy-first approach means we never have to worry about compliance.”
VP of Engineering
Social Content Platform
+45% pages/session
“Integration took 2 hours. Our article recommendations went from generic to genuinely personalized. Readers are exploring more content than ever.”
Head of Product
Digital News Publisher
+27% watch time
“Our homepage used to open on whatever was newest. Now it opens on what this particular viewer is likely to finish, and session length went up without us adding a single new title.”
Head of Product
Streaming Platform
Nothing to leak,
because nothing is kept.
Pulse doesn't need to know who someone is to know what they want. That isn't a policy we wrote, it's how the system is built, which means compliance review tends to be short.
Zero personal data
No names, emails, phone numbers, addresses, or payment details are collected, requested, or stored. There is no user table, because there are no users — only sessions.
No cookies, first or third party
Nothing is written to the browser and nothing needs consent between visits. Cookie-consent changes and browser tracking restrictions don't degrade what Pulse returns.
GDPR and Decree 13 aligned
Designed against GDPR, CCPA, and Vietnam's Decree 13/2023 on personal data protection. Data residency options in Singapore, Frankfurt, and Virginia. Full DPA available on request.
See it read your traffic.
Twenty minutes, your storefront, live signals. You'll see the intent profile Pulse builds for real sessions on your own site — before you decide anything.
