SalesPilot
success story

For brands and online stores

One recommendation intelligence.

A whole ecosystem of possibilities.

SalesPilot decides which product to show at every moment. CrossUp is the success story of applying it well in online stores.

crossup.ai

CrossUp · the problem

Every visit is already paid for. Every purchase has to be worth more.

$

Acquiring shoppers keeps getting pricier

Meta and Google ad costs rise year after year.

=

Average order value stays flat

Shoppers only buy what they came for.

↑

CrossUp raises what every visit is worth

More products per order, same ad spend.

Ad spendunchanged
$100.000
Saleshigher AOV
+9,2 %
$310.000$338.500
ROASsales per dollar of ad spend
3.1x3.4x

Illustrative example using the +9.2% AOV lift measured in stores running CrossUp.

SalesPilot · CrossUp

SalesPilot

Recommending is a game of Battleship

Each square crosses a shopper with a moment
The ships are the right product
The color shows where it aims next
First shots1 of 4
After learning5 of 6

Every shot, hit or miss, teaches SalesPilot where to aim the next one.

SalesPilot · CrossUpIllustrative example

SalesPilot in motion

Data, moments and results, connected

Data sources
Buying moments
Users
Orders
Products
Categories
Interactions
History
Oversized tee
Sheet set
Water bottle
Padel racket
Skincare
15 kg dog food
BT earbuds
Puffer jacket
Scented candle
Sneakers
Travel mug
Foundation
SalesPilot
engine
the inside
Output
Product
Add to cart
Cart
Checkout start
Checkout
Purchase page
see the engine
Product
Add to cart
Cart
Checkout start
Checkout
Purchase page
Add to cart
Black oversized tee
Matching cap
+$12.40extra
Accepted· +$12.40 extra
Tee ↔ Cap affinity
Product
Queen sheet set
Pillowcases
+$8.90extra
Accepted· +$8.90 extra
Home textiles: bundle wins
Cart
1 L steel water bottle
Sleeve + brush
+$15.60extra
Accepted· +$15.60 extra
Bottle ↔ Sleeve gets stronger
Checkout
Vertex padel racket
Grip + protector
+$6.30extra
Ignored· no change
Checkout: 1 item, not 2
Checkout start
Skincare routine
SPF 50 sunscreen
+$11.20extra
Accepted· +$11.20 extra
Routine → SPF climbs
Purchase page
15 kg dog food
Dental chews x3
+$4.80extra
Accepted· +$4.80 extra
Repeat at 30 days climbs
Add to cart
In-ear earbuds
Hard case
+$5.40extra
Accepted· +$5.40 extra
Audio → Case climbs
Product
Puffer jacket
Wool scarf
+$9.70extra
Ignored· no change
Scarf drops on the page
Cart
Vanilla scented candle
Reed diffuser
+$7.10extra
Accepted· +$7.10 extra
Scents: cross-sell climbs
Checkout start
Running sneakers
Compression socks
+$6.90extra
Accepted· +$6.90 extra
Returning: ticket climbs
Checkout
Insulated travel mug
Coffee + caddy
+$10.30extra
Ignored· no change
Checkout: drop the 2nd item
Purchase page
Liquid foundation
Sponge + brush
+$5.90extra
Accepted· +$5.90 extra
Makeup → Applicators
Learns from the outcome
1

Understands the store

Products, categories, orders and what every shopper does.

2

Decides what to show

Valid complements, in stock and within the merchant's rules.

3

Picks the moment

Product page, cart, checkout or after the purchase.

4

Learns from the result

Every accept or reject improves the next decision.

SalesPilot · CrossUpLive product

The everyday decision

The catalog is the same.
The decision changes with the moment.

Orange blazer
Orange BlazerViewing
Black turtleneck
Black TurtleneckCompletes the look
Palazzo pants
Palazzo PantsCompletes the look
Gold hoops
Gold Hoops
Thin bracelet
Thin Bracelet
Burgundy sweater
Burgundy Sweater
Product pageWhat helps close the sale?
Cart close to free shippingWhat adds up without eating the margin?
After the purchaseWhat to offer for next time?
SalesPilot · CrossUpContext

SalesPilot

How SalesPilot decides, on every visit

1Reads the moment
Orange blazer in the cartOrange Blazer in the cart
Size MR$ 40 away from free shipping
2Filters the catalog
Out of stock
Doesn't match
No size M
Other category
Low margin
3Ranks by value
1
Black TurtleneckCompletes the look
2
Palazzo PantsBought together
3
Gold HoopsLightweight, unlocks free shipping
4Shows and learns
Complete the look
Turtleneck
Palazzo
Hoops
Took the turtleneck · +R$ 109.50

Every response improves the next decision

SalesPilot · CrossUpLive product

Engine and application

SalesPilot is the engine. CrossUp is the success story of applying it well.

CrossUpSuccess story · in online stores
E-commerce platformsWhat's next
Ecosystem appsWhat's next
SalesPilot
Decides which product to show, under which rules and when

SalesPilot decides

A recommendation intelligence that can power any surface: storefront, checkout, email, chat or app.

CrossUp proves it

The app that turned that intelligence into incremental sales, with the merchant in control.

SalesPilot · CrossUpLive product

CrossUp · success story

Six shopping moments, one intelligence

Offer on the product page
1Product page
Add-to-cart popup
2On add to cart
Recommendation carousel in the cart
3In the cart
Recommendations at checkout start
4At checkout start
Last chance inside checkout
5Inside checkout
Offer on the confirmation page
6Order confirmation
SalesPilot · CrossUpLive product

CrossUp in action

This is how it looks in a real store

Frame from the CrossUp video

The video walks through each moment with real products: product page, cart, checkout and confirmation.

And the interactive simulation lets you try the real widgets on a demo store.

Open the simulation app.crossup.ai/simulacion
SalesPilot · CrossUpLive product

Every vertical, its own logic

A different recommendation for every store

Dog harness+Leash
Harness → leashSame style
Protein+Shaker
Protein → shakerUsed together
Diffuser+Refill reeds
Diffuser → reedsRefill
Sneakers+Insoles
Sneakers → insolesComplement
Cream+Serum
Cream → serumFull routine
Ring+Necklace
Ring → necklaceSame style
SalesPilot · CrossUpIllustrative examples

What CrossUp has already proven

More value in every order, without friction

+38 %Beauty WebAverage order value · Apr–Jul 2025
+51 %FungiAverage order value · Apr–Jul 2025
+27 %ÍntegraAverage order value · Apr–Jul 2025
+9,2 %

AOV in stores running CrossUp, since May 2025

4,6 ★

average merchant rating, 261 reviews

Why it worksThe merchant is in chargeNative integrationNo valid offer, no widgetEvery sale tied to its order
SalesPilot · CrossUpResults published on crossup.ai

Where commerce is headed

From a widget to the platform

Today
You may also like
One block, on one page.
One engine for every store
A section in every theme
A feed for every shopper
Which leash goes with this harness?Print Leash · R$ 79.50
A chat that sells
A gift for my mom
Ready for AI agents
from home page to post-purchase
At all 8 moments
Where the most value is waiting:the cartcheckoutfree shipping
SalesPilot · CrossUpVision

SalesPilot as an agent

An agent you can talk to that connects to everything

Your store adminView storeHelp
HomeSalesProductsCustomersMarketingDiscountsAnalytics
Marketing
Active campaigns
Combos and promotions
In-store recommendations
AI assistant · uses SalesPilot via MCP
Build me combos for Children's Day
Based on what's bought together in your store and the available stock, I suggest these three:
+
First toysBought together · ages 2 to 4
Publish
+
Learn through playSame age range · in stock
Publish
+
Afternoon outdoorsGood margin · light shipping
Publish
Ask your assistant anything
SalesPilot Agent
What should Moda Sur recommend for Hot Sale?
Three things: add "Complete the look" to the cart, build the Blazer + Turtleneck combo and pause the out-of-stock line. Should I apply it?
ApplySee why
Connects to any app Open platform · in development
MCP

Any AI agent uses it as a tool.

recommend_products(store, context)
API

One decision per request, from any system.

POST /v1/decisions
SDK

For themes, checkout and ecosystem apps.

sp.recommend({ surface: "cart" })
CLI

For teams and agencies that automate.

salespilot combos --store moda-sur
SalesPilot · CrossUp

Inside SalesPilot · the catalog

From a loose catalog to a map of relationships

Similar isn't complementary. Two pairs of pants compete; pants and sneakers go together.

Families that look alike and compete
Bridges that are bought together

In a real store, a decision model evaluated every pair in the catalog:

39products
741pairs evaluated
25 minend to end
US$0.02total cost
SalesPilot · CrossUpReal store · offline evaluation

Inside SalesPilot · the architecture

From thousands of products to the three that matter

~5.000products in the catalog
~100candidates
12ranked by value
3on screen, in milliseconds
Understands the store
01
Storefront script

Installs in the store and shows the offer at six moments.

Platform SDK · 6 moments
02
Signals

Catalog, stock, orders and browsing, in real time.

real-time events
03
Embeddings

Every product is a vector: similar items sit close together.

text-embedding-3-large
Decides the offer
04
Candidates

What sells together and what looks alike, in parallel.

co-purchase with lift + similar items
05
Ranking

Ranks by expected value. If nothing adds value, it waits.

two-tower · expected value
06
Quality Review

Complement, substitute or variant: only complements pass.

typed decision model
Measures and learns
07
Optimizer

A control group isolates what SalesPilot adds.

holdout · Bayesian readout
08
Agentic SalesPilot

Agents that study each store and build its strategy.

agents · Cold Start
09
Your rules

The merchant's picks always take priority.

merchant rules
In productionIn validation, on the way to productionWorks outside the live path
SalesPilot · CrossUp

The complete journey

One intelligence, from the first visit to the next purchase

Discover

Comes back to the store

An email or the app shows what they're interested in.

Email · App
Explore

Finds what matches

The product page completes the look.

Store theme
Decide

Fills the cart

And unlocks free shipping with the right product.

Cart · Free shipping
Ask

Asks on WhatsApp

Gets real options, in stock.

Chat · WhatsApp
Pay

Adds one more thing at the end

Without slowing payment.

Checkout
Return

Repeat purchase

At the right time, with what comes next.

App · Email
SalesPilot

The same decision at every touchpoint: which product, for whom and when.

The merchantsets the rules from their admin
SalesPilot · CrossUpcrossup.ai