Ms Weckert's algorithm doesn't lie. That's Coles' problem.

For years, the dominant promotional strategy at Coles and Woolworths has been a three-step manoeuvre:

1. Hold a product at its normal price for several months
2. Spike the price sharply for a few weeks
3. Drop it slightly below the spike and apply a "Prices Dropped" or "X% off" label

The mechanic depends entirely on consumer memory failure. If you remember last week's price, the trick doesn't work. But most shoppers don't track prices week to week. They see a discount label and assume they're getting a deal.

An agent has no memory failure. An agent has a ledger.

We know this mechanic isn't theoretical because both Coles and Woolworths are currently in Federal Court facing exactly this allegation. The ACCC found specific products where the "regular" reference price was an artificial spike, not the actual long-run price, and the resulting "discount" was higher than what consumers had been paying for months.

We ran the same pattern detection (a mode price held for six or more weeks, followed by a spike of 20%+, followed by a "dropped" price that stays above the mode) against 52 weeks of our own pricing data across 74,000+ products. We found 148 active examples.

Here's the pattern, illustrated with one product that has become something of a case study.

Across our full dataset:

The supplements category is the most extreme. Coles Own B3 500mg tablets: normal price $12.25 for 11 weeks, spiked to $29.50, "dropped" to $22.13. The label says 25% off. The actual increase from what you were paying is +81%.

Ms Weckert described buying agents that could "absorb much more information about any particular product than an overwhelmed human customer can." She's correct. That's precisely the problem.

The billions of dollars worth of promotional mechanics that Coles and Woolworths have built -- the in-store theatre, the red labels, the percentage callouts: all optimised for a customer who doesn't have a spreadsheet in their pocket. Agents are spreadsheets.

Ms Weckert frames agentic commerce as a future shift. But the core capability she's describing, absorbing more price information than an overwhelmed shopper can hold in their head, already exists. It's just not being deployed by the retailers.

Pinch tracks 74,000+ products across Coles, Woolworths, ALDI and Harris Farm every week. It stores price history. It detects hi-lo cycles. It normalises unit prices so you're comparing grams to grams, not pack sizes to pack sizes. It does exactly what an agent would do, except it's in your pocket right now, not on a retailer roadmap for 2030.

The difference is who it works for.

A Coles agent, or a Woolworths agent, will optimise within that retailer's catalogue. It will route you to the products that are best for the platform's economics: promoted lines, private label, items with supplier subsidies behind them. That is not conspiracy; it is the stated commercial model.

An independent tool compares across retailers. It has no supplier deals, no trade spend, no promotional calendar to protect. When ALDI's tuna is 40% cheaper per unit than Coles' "discounted" tuna, it says so. When Harris Farm's milk is a dollar less than the duopoly price, it says so. When a "Prices Dropped" label is hiding a net increase, it flags it.

The hi-lo pricing playbook documented on our Hi-Lo Pricing page (14,473 products with regular price cycles, 2,308 where the "sale" price is the new floor) is the kind of structural pattern that an individual shopper has no chance of detecting on their own. You would need to track thousands of prices over months to see it. That is precisely what Pinch does.

The Illusory Discounts we document (230 products currently on "special" at prices higher than what you were paying months ago) represent the output of a pricing system that is designed to overwhelm. It relies on the gap between what the label says and what the price history shows. Close that gap and the mechanic stops working.

That is what Pinch does. Not in 2030. Today.

Earlier this year, CHOICE ran its quarterly basket survey across the major supermarkets. We matched those 17 products against our own pricing data to see how the retailers compare when unit prices, not pack sizes, are the metric.

When sizes are normalised, the picture shifts. By our calculation, ALDI wins 9 of 17 items on unit price: carrots, chicken, baby cucumbers, blueberries, yoghurt, ham, sultanas, wholemeal bread, and tuna. Harris Farm wins 3 (milk, strawberries, Vegemite). Coles wins 2 (Weet-Bix, cheese slices).

An agent doing a weekly grocery shop optimised for lowest unit price would split shop between ALDI and, depending on the week, one or two others. It would not default to Coles. It would not default to Woolworths.

There is one more dimension that doesn't appear in Ms Weckert's speech, perhaps for obvious reasons.

From 1 July 2026 (seven weeks from today) the Supermarkets Excessive Pricing Prohibition comes into force. It targets retailers with annual revenue exceeding $30 billion from Australian supermarket operations.

Exactly two companies clear that threshold: Coles and Woolworths.

ALDI is not covered. Harris Farm is not covered. IGA is not covered. Independent butchers, greengrocers, and specialty retailers are not covered.

The incoming regulations also strengthen unit pricing rules specifically to target shrinkflation (the practice of reducing pack sizes while holding prices). Agents, it should be noted, are exceptionally good at detecting shrinkflation.

So from July: the retailers best positioned in an agentic future (consistent pricing, competitive unit costs, no hi-lo playbook) are also the ones operating outside the scope of the new compliance regime. That is not a coincidence. It is a market structure question the ACCC has been asking for two years.

Ms Weckert is right that agentic commerce will transform the grocery sector. Her prediction of 30% of online orders handled by agents by 2030 is, if anything, conservative given current AI adoption curves.

But the transformation she describes is one that rewards price consistency and punishes complexity. Hi-lo pricing cycles, illusory discount labels, coordinated category spikes: these are all mechanisms optimised for a customer who is paying some attention but not full attention. Agents pay full attention, all the time, with perfect recall.

The retailers entering this era with the most to lose are the ones with the most elaborate promotional mechanics. The ones with the most to gain are the ones who have been pricing simply and consistently all along.

The question for Australian shoppers is not whether agents will change grocery. They will. The question is whether you wait for Coles and Woolworths to build agents that serve their interests, or whether you use an independent tool that serves yours.

Pinch exists because the gap between what a price label says and what the data shows is too wide for any individual shopper to close on their own. We close it with 65,000 products, four retailers, and a weekly price history that no promotional calendar can hide from.