Book a call

Services · S1 to S4

Four ways in

Each offer is built around a number you already report on. Start with the diagnostic, or go straight to the build you know you need.

Every service ends in something measurable: a business case, a model in production, or a score you can track.

S1 · Fixed-scope sprint

AI Commerce Diagnostic

Find the three AI plays worth funding in your funnel, with the numbers to defend them.

Timeline
3 to 4 weeks
Model
Fixed-scope sprint
You get
Opportunity map · Ranked use-case backlog · Business case per top-3 play · 90-day roadmap · Executive readout
Book a call about S1

The problem

Every vendor has an AI pitch and every team has a pilot. Very few organizations can say which AI investment will move revenue first, what data it needs, and what it will cost to run.

What we do

  • Funnel and data audit across commerce, CRM and engagement platforms
  • Revenue-leak sizing for each stage, modelled on your numbers
  • Use-case shortlist scored on value, feasibility, risk and time to proof
  • Architecture sketch and a build-versus-buy view per use case

Under the hood

A funnel model built on your own data. Leaks sized stage by stage. Use cases scored on value, feasibility, risk and time to proof. Build or buy, play by play.

The outcome

A board-ready investment case and a sequenced roadmap you can act on the following Monday.

Questions

Do we need clean data before a diagnostic?

No. Assessing data readiness is part of the work. Most clients learn their data is better for some use cases than they feared, and worse for others than they hoped.

What happens after the diagnostic?

You own the roadmap. Many clients continue into a build engagement for the first use case, but there is no obligation.

S2 · Build engagement

Agent-Ready Commerce

Make your catalog and checkout legible to the AI agents that now shop for your customers.

Timeline
6 to 10 weeks
Model
Build engagement
You get
Catalog quality scorecard · Enrichment pipeline · Agent-readiness checklist · Channel and protocol recommendation
Book a call about S2

The problem

When a shopper delegates a purchase to an assistant, the agent reads structured data, not your homepage. Thin attributes, stale availability and unclear policies mean your products are simply not chosen.

What we do

  • Catalog enrichment at scale: attributes, descriptions, alt text, translations
  • Feed and API readiness for agents (price, stock, shipping, returns, policies)
  • Protocol and channel assessment (ACP, AP2, marketplace and assistant programmes)
  • Design for how your brand shows up inside assistants

Under the hood

Attribute extraction and enrichment with language models, plus human review. Product feeds agents can read: price, stock, shipping, returns. Protocol fit across ACP, UCP and AP2.

The outcome

Products that agents can find, compare and buy, and a catalog that converts better for humans too.

Questions

Is agentic commerce real yet or still hype?

Both. Protocols and early checkouts exist and are changing quickly. The no-regret move is the same either way: structured, accurate, machine-readable product data. It pays off in search and on-site conversion today.

Do we have to pick ACP or AP2?

Not yet. They solve different layers (checkout versus payment authorization). We help you get ready for both without betting on one.

S3 · Build engagement

Revenue Intelligence

Predict who will buy more, who will leave and who is gaming you, then act on it in your CRM.

Timeline
8 to 14 weeks
Model
Build engagement
You get
Production models · Scoring pipeline · CRM/engagement integration · Experiment design · Model monitoring
Book a call about S3

The problem

Most retention and recovery programmes treat every customer the same. Discounts go to people who would have bought anyway, high-value accounts drift away unnoticed, and abuse hides inside honest volume.

What we do

  • Propensity, uplift and churn models trained on your own history
  • Account health and revenue-at-risk scoring for B2B teams
  • Next-best-action and message generation wired into Salesforce, Braze, Klaviyo or similar
  • Measurement design with holdouts, so results are provable

Under the hood

Propensity, uplift and churn models in classical machine learning. Generative AI writes the message and never makes the decision. Delivered through your CRM, measured against a holdout.

The outcome

Margin spent only where it changes behaviour, and a measured lift you can show finance.

Questions

Is this generative AI?

Mostly not. The targeting brain is mature, well-understood machine learning. Generative AI is the layer that writes tailored messages and briefings on top of it.

Will this work in Salesforce and Agentforce?

Yes. Scores and recommendations surface where your teams already work, and Agentforce can act on them. We are platform agnostic where you are not on Salesforce.

S4 · Build engagement

Experience & Compliance Automation

Remove the friction and the legal exposure from accessibility and cross-border selling.

Timeline
6 to 12 weeks
Model
Build engagement
You get
Compliance baseline · Remediation pipeline · Landed-cost integration · Monitoring dashboard
Book a call about S4

The problem

Accessibility gaps and surprise duties are both friction and risk. They lose customers quietly and create legal and operational exposure loudly.

What we do

  • Automated accessibility scanning and remediation (alt text, captions, markup)
  • Ongoing compliance scoring and regression monitoring
  • Landed-cost and duty calculation at checkout, HS classification assistance
  • Clear separation of AI interpretation from deterministic, auditable calculation

Under the hood

Automated accessibility scanning with human review. Landed cost from official tables in a rules engine. AI suggests classifications, never the rate.

The outcome

A measurable compliance score, fewer refused deliveries and new markets opened safely.

Questions

Can AI make a site fully accessible on its own?

No, and anyone who promises that is selling an overlay. AI removes the bulk of repetitive remediation; people still review, test with assistive technology and own the result.

Will AI calculate our duties?

AI suggests classifications and reads regulations. A rules engine calculates the duty from official tables. The model never invents a rate.

Next step

Bring one number you want to move.

Thirty minutes with the founder. You'll leave with a straight answer on whether AI is the right tool for it.

Book a call

AI is… the Future.