The AiESL Pricing Agent weighs demand, stock, margin targets and competitor signals, then recommends or automatically applies the right shelf price and pushes it to every connected label in under a second — with floor prices, approval rules and a full audit trail.
Demand, stock and competitor prices change daily; paper-based repricing runs weekly. The decision is ready long before the shelf is.
Fresh stock is marked down at the end of its life, in one big step — or not at all — so margin and product are both lost.
Category teams compare competitor prices by hand and re-tag in batches, so reactions come days after the market moved.
The AiESL Pricing Agent is an AI decision engine that weighs demand, stock, margin targets and competitor signals, then recommends or automatically applies the shelf price — and pushes approved prices to every connected label. It combines large language models with retrieval-augmented generation (RAG), runs in a cloud-edge architecture and responds in about 350 ms; approved prices propagate store-wide in under a second.
| Step | What happens | Your control |
|---|---|---|
| 1. Sense | Ingests sales and demand from the POS, stock and expiry from ERP/WMS, and competitor price signals. | Choose which data sources and categories are in scope. |
| 2. Decide | Computes a price inside your floor prices and margin limits. | Set objectives per category: margin, waste, competitiveness. |
| 3. Approve | Applies automatically under auto-approval rules, or waits for a person. | Choose recommend-only or auto-apply. |
| 4. Apply | Pushes the price to every matching label, any connected brand. | Target by store or promotion. |
| 5. Measure | Tracks sales, margin and waste after the change and feeds the result back. | Full audit trail of every recommendation and change. |
The mechanics of the loop and how to prove its ROI are explained in depth in AI dynamic pricing with ESL: how it works. On our five-level AI ESL scale, this module operates at Levels 3 and 4.
| Format | Typical objective | What changes on the shelf |
|---|---|---|
| Supermarket & grocery | Reduce waste on perishables | Time- and demand-based markdowns on fresh stock, inside margin guardrails, instead of one late cut. |
| Electronics retail | Stay competitive without a price war | Prices react to competitor signals automatically, never below your floor price. |
| Convenience | Fast, local promotions | Instant promo swaps by store or time window, without re-tagging. |
| Multi-store chains | Consistent head-office pricing | Head office pushes a price or promotion to every store at once; every label confirms the update. |
Format pages: supermarket, electronics, convenience.
Every recommendation is bounded by rules the engine cannot cross, and every change is logged. That is what makes automated pricing governable — and defensible to shoppers and regulators.
| Guardrail | What it does |
|---|---|
| Floor prices and margin limits | Hard boundaries per product or category; no recommendation can breach them. |
| Approval modes | Recommend-only, or auto-apply under auto-approval rules — set per category. |
| Audit trail | Who or what changed each price, when, why and with which inputs. |
| Same price for every shopper | The engine works on product, stock and market data, not on individual shoppers’ data; everyone sees the same shelf price. |
Regulation is moving in this direction: New Jersey’s 2026 Fair Price Protection Act bans individualised pricing based on consumer data. What the research says about ESLs and surge pricing — and what it says about markdowns — is summarised in are ESL labor savings overstated?
| Question | Answer |
|---|---|
| Which labels can it drive? | Any label connected to the AiESL open middleware — 5+ brand integrations, including labels already on your shelves where the brand is supported. |
| Cloud or on-premise? | As a module of the managed cloud service, or as the price-strategy engine in the on-premise full platform. |
| We already have a pricing engine. | Keep it. Send its prices to the middleware API and use only the delivery, confirmation and audit layers. |
| What data does it need? | Product master and prices from ERP/POS, sales history, stock levels and, for perishables, expiry or receiving dates. Competitor data is optional. |
We do not publish uplift percentages; we would rather measure yours.
Start with one or two fresh or competitive categories in a few stores, with matched control stores.
Run the engine in recommend-only mode for several weeks so category managers can review every proposal.
Switch on auto-approval for the rules that proved themselves, and measure margin and waste against control stores.
Talk to a solution engineer about your stores and systems — no sales script.
Talk to an engineerIt is a pricing engine that uses demand, stock, margin and competitor data to recommend or automatically set shelf prices, and electronic shelf labels that display the new price within seconds. Without electronic labels the decision cannot reach the shelf fast enough; without the engine, the labels only show prices someone typed in.
The Pricing Agent responds in about 350 ms in its cloud-edge architecture, and approved prices propagate to every connected label store-wide in under a second. In practice, how often prices change is set by your rules, not by the speed of the system.
Yes. Each category can run in recommend-only mode, where a person approves every change, or auto-apply only within the auto-approval rules you define. Floor prices and margin limits are hard boundaries, and every recommendation and change is recorded in an audit trail.
Not in how retailers use it in practice, and not in what the research shows. The best evidence links electronic shelf labels to smaller, more frequent and mostly downward price changes on perishables, and a study of more than 180 million price observations found no ESL-driven surge pricing.
Not if your label brand is one of the 5+ brands the AiESL middleware supports. The Pricing Agent sends prices through the middleware, which drives each brand’s labels through its own gateways.
Yes. Keep your engine and send its output to the middleware API; you still get delivery to any connected label, update confirmation and an audit trail.
Start in recommend-only mode on a few stores and measure margin and waste against control stores.
Get a Custom Quote