How EcoWise AI works

Till data in. Advice out.

Five steps take a month of till transactions and turn them into a plain-English action plan for your shop. Scroll to follow the journey.

Follow the journey ↓ Join the pilot

Step 1 · The upload

It starts with your till

Every transaction your shop rings up, uploaded securely. The first time it's two years of history; after that, a monthly drop — no new hardware, nothing to install.

12,431 transactions in a typical first upload

Step 2 · The dictionary

Chaos becomes order

Till data is messy — the same product hides behind a dozen different names. Every SKU is cleaned, categorised and given a role in your range. It happens once, at onboarding, with AI help on new products ever after.

Step 3 · Three tiers

Three tiers of intelligence

Your ordered data feeds three layers of analysis, each refreshing on its own clock:

  • Performance — the descriptive view: value, units, year-on-year, category mix.
  • ML patterns — models surface the seasonal trends and basket clusters hiding in your transactions.
  • Context — market trends, competitor offers, new products and layout strategy, kept fresh.

Step 4 · The AI core

The AI reads everything

All three tiers feed one recommendation engine. It weighs what's selling, what's shifting and what the market is doing — then decides what deserves your attention this month.

Step 5 · Your report

And writes your report

No dashboards and no jargon — a short, plain-English report specific to your shop, with a reason attached to every recommendation.

Your soft drinks chiller · 3 recommendations This month · EcoWise AI
Fanta Orange 500mlDelist

Sold 4 units last month and holds a full chiller facing that could earn three times more.

Pepsi Max 500mlAdd

The best-selling soft drink in shops like yours, and it is not on your shelf.

Red Bull 250mlPromote

Already sells well at full price. Pairing it with a snack meal-deal should lift it further.

Projected impact: +11% category sales over 4 weeks

Illustrative example — your report is built from your own till data.

The detail

Each layer runs on its own clock

One secure upload feeds three tiers of intelligence. Every layer refreshes on its own cadence, so your advice stays current without you lifting a finger.

1

Secure upload

You log in with your store code and 2FA, and share your till data. First time it's two years of history; after that, a monthly drop.

Monthly
2

Category dictionary

Every SKU is cleaned, categorised and given a role. Chaos becomes order — once, at onboarding, with AI help on new products.

One-time
3a

Performance tier

Python aggregation builds your descriptive view: value, units, year-on-year, category mix.

Monthly
3b

ML tier

Machine-learning models find seasonal trends and basket clusters in your transactions.

Every 4–6 months
3c

Context tier

Macro trends, competitor offers, NPD, best sellers, meal deals and layout strategies — kept fresh.

Monthly
4

AI recommendation layer

The LLM reads all three tiers and writes range, promotion and price recommendations — each with a reason.

Monthly
5

You act, it learns

Tick off the recommendations you implement. The system measures the exact impact of your actions and gets sharper.

Continuous

Join the pilot

See what it would say about your shop

We're piloting with independent stores through the NFSP — free during the pilot, no contract, and you can leave at any time.

Join the pilot Read a sample report