Make agents
understand
your catalog
to convert.

Toffee is the intelligence layer between your catalog and agentic traffic — so AI shopping agents understand your brand, recommend the right products, and convert through your stack.

01
Stylist conversation
Your AI stylist understands occasion, vibe, and fit.
Message Toffee Stylist…
02
Brand-aware retrieval
Toffee
Toffee
Toffee surfaces products with the brand context your request needs.
AURÉLIE
Boutique
Resort luxe
NOMAD
Resort
Bohemian
VERDE
atelier
Minimal
LUNA
EDIT
Editorial
Brand-coherent · Catalog context
The market is moving

Agentic commerce is happening now.

Surfaces, protocols, and checkout rails are shipping fast. Being reachable is not the same as being understood, selected, and converted.

393%.
AI traffic to U.S. retail, year over year.
Adobe, Q1 2026
+42%.
Lift in conversion from AI-referred sessions.
Adobe, March 2026
UCP
Google ships an agentic commerce protocol.
Google, 2026
$262B
Agent-influenced global online spend, 2025 holiday.
Salesforce, 2025
The problem

A product feed can list your catalog.
It cannot explain your brand.

01

Brand identity flattens

Distinct brands cannot be reduced to title, image, size, color, price.

02

Agents miss meaning

A feed shows what a product is, not why it belongs in a look or a moment.

03

Exposure does not compound

Without owned traces, agent visits stay traffic, not memory or learning.

What agents see today

Flat feed

Title. Price. Image. Color. Size.

SKU-1000
SKU-1013
SKU-1026
SKU-1039
SKU-1052
SKU-1065
SKU-1078
SKU-1091
SKU-1104
Reachable
What Toffee delivers

Brand commerce graph

Meaning. Relationships. Policy. Memory.

brandstyleoccasioncomplementssubstitutespolicycustomer ctxmemoryactions
Understandable
The graph

From static product data
to agent-ready meaning.

Toffee builds a merchant-specific Brand Commerce Graph. Clusters of nodes and edges that encode what your products mean, how they relate, and what agents are allowed to do with them.

EDGES
complementssubstitutesbelongs-to-lookconflicts-withgood-for-occasioneligible-for-checkoutpreferred-for-profileconverted-after
cluster

Products and variants

SKUvariantsmediapricing
cluster

Brand identity

voicetieraestheticpositioning
cluster

Style universe

archetypeoccasionmoodfit
cluster

Relationships

complementssubstitutesbelongs-to
cluster

Policies and actions

promotionsreturnseligibility
cluster

Customer context

profilehistoryloyaltyconsent
cluster

Inventory and paths

stockfulfilmentcheckout
cluster

Merchant memory

tracessessionsoutcomes
Architecture

One graph. One runtime.
Continuous optimization

Layer 01

Brand Commerce Graph

Persistent intelligence. Meaning, brand, relationships, policies, actions, memory, outcomes. The merchant's source of truth for agents.

persistentmerchant-ownedqueryablegoverned
Layer 02

Agentic Conversion Runtime

Live decisioning. Retrieval, ranking, permissioning, and routing. Every agent request is shaped by your graph, then routed through your stack.

retrieve
graph query
rank
brand, intent, policy
permit
action policy
route
merchant rail

The graph creates understanding. The runtime turns understanding into action and conversion.

How it works

Continuous
representation
learning.

Every agent interaction sharpens the next one. Traces from real shopping journeys feed back into the graph as merchant-owned context. The longer Toffee runs, the better your representation gets.

01IntegratePIM, DAM, ERP, CRMs
02Synthesizebuild merchant hypergraph
03Exposeprotocols, scoped APIs
04Optimizeclassify, rank, frame
05Routemerchant-preferred rail
06Learnupdate the hypergraph
continuousrepresentationlearning
STEP 1
Integrate
PIM, DAM, ERP, CRMs
STEP 2
Synthesize
build merchant hypergraph
STEP 3
Expose
protocols, scoped APIs
STEP 4
Optimize
classify, rank, frame
STEP 5
Route
merchant-preferred rail
STEP 6
Learn
update the hypergraph
Where Toffee sits

Where Toffee sits
in the agentic commerce stack.

Surfaces and protocols give agents access to merchants. Toffee is the layer that turns brand logic, product meaning, and conversion strategy into something agents can actually understand and act on.

LAYER 01
Consumer surfaces
ChatGPT, Claude, Gemini, Perplexity, Operator. via UCP, ACP, MCP.
LAYER 02
Rails
Stripe, ACP, UCP, PayPal, merchant checkout, PSPs.
LAYER 03
Toffee.
Brand Commerce Graph. Memory. Action policy. Conversion optimization.
LAYER 04
Merchant systems
catalog, inventory, pricing, cart, checkout, loyalty.

Toffee sits directly above the merchant stack, effectively integrated into it. Brand logic, policy, and memory travel with every agent request.

The agent journey

From a single prompt
to a merchant-owned purchase.

One scenario, six frames. Toffee is the layer that turns a vague intent into a coherent, on-brand, converting outcome.

FRAME 01
Shopper intent
"Build me a complete outfit for a gallery opening under €700."
FRAME 02
Agent request
ChatGPT calls the merchant via UCP, asks for an on-brand look.
FRAME 03
Toffee graph response
Brand, style, occasion, policy, stock, all considered together.
FRAME 04
Coherent recommendation
A real outfit, not a list of substrings, ranked by intent and brand fit.
FRAME 05
Branded checkout
Routed through the merchant’s preferred rail with full context.
FRAME 06
Trace returned
Outcome flows back to the graph as merchant-owned learning.
Where it lands first

Built for categories where
meaning drives conversion.

Fashion.

Distinct brand worlds preserved, not flattened into a generic catalog.

brand contrasteditorial intentseasonality

Luxury.

Outfit coherence and meaning matter more than keyword match.

curationtiertrust

Marketplaces.

Many brands, policies, and styles under one merchant surface.

multi-brandpolicy graphtaxonomy
What merchants get

Outcomes that compound,
not traffic that disappears.

01

Brand-preserving representation

Products live inside your brand world, not as flat SKUs.

02

Higher-quality conversion

Optimize how agents rank, frame, act, and route.

03

Owned learning loop

Every interaction becomes reusable merchant context.

04

Protocol and rail flexibility

Any rail, any protocol, any future surface.

05

Stack stays in place

Adapts your merchant layer. No replatforming.

06

Marketplace-ready

Many brands and styles under one surface.

Implementation

Start lightweight. Go deeper
where conversion demands it.

L1

Agent readiness

Structured catalog. Brand and policy graph. Agent-readable surface.

structured catalog
policy graph
agent endpoint
L2

Action layer

Cart. Checkout. Permissions. Trust tiers. Routing.

cart actions
permissions
rail routing
trust tiers
L3

Optimization layer.

Memory. Traces. Ranking. Brand representation tuning.

traces
ranking
brand rep tuning
attribution
Built to fit

Designed for the systems
merchants already run.

Stack-agnostic

PIMs, feeds, APIs, checkout.

Rail-compatible

any rail, any PSP.

Owned memory

merchant-controlled context.

Brand-safe

structured, not scraped.

Marketplace-ready

many brands, one surface.

FAQ

Quick answers.

No. Toffee routes through your existing checkout and rails. Stripe, ACP, UCP, PayPal, your own PSP. Conversion stays on your stack.

Make your catalog agent-ready
before agents define it for you.

Toffee gives merchants control over how AI agents understand, represent, act on, and convert through their brand.

Protocols make you reachable.
Toffee makes you worth choosing.

Enter demo Read the brief