What needs you today, ranked by money.
marql opens on Today: one queue with what your AI employees prepared and what marql found in your data, the biggest benefit first. You decide. marql checks it was done and measures what it gave.
TodayFri · Demo network · 12 stores
Proven, 30 days
3 120 €
Results: 4
Waiting for you
7
largest ≈740 € per week
Revenue, 7 days
184 300 €
↗ 2.1% vs previous 7 days
Week goal
2 140 / 3 000 €
Updated 11:40 · data from 11 of 12 stores · how it is counted
Decisions · 7by benefit
marql · whole network
Review prices and suppliers for low-margin dairy SKUs
Why now: dairy margin fell 4.2 pp in five stores in three weeks while the rest of the network held.
Saves ≈740 € · per week · estimate
marql · whole network
Clear dead stock and release tied-up capital
≈6 400 € one-off
Ana · AI supply employee · AI
Transfer · North Park → City Center
Saves ≈210 € per week
Ana · AI supply employee · AI
Supplier order · dairy supplier → Central warehouse
Benefit not estimated · Cost: purchase 1 180 €
3 more awaiting review ⌄
What changed since yesterday
Yesterday 26 900 €, +3.4% against the same day a week earlier.
New decisions: 2.
Result proven: Move slow stock to two stores, 410 €.
StoresOpen Live
2 of 12 behind pace · 860 € short by 11:40
Most of what marql finds never reaches you.
One night on a 12-store demo network: what the detectors found, what the filter kept off your list, and what was left for the morning.
41
findings last night · 12 stores
Found in your data
Detectors read every store against its own normal, and your AI employees make their morning pass from 06:00 local time. Connectors bring data in every few minutes and the full scans run nightly. We publish the intervals instead of calling it real time.
−34
kept off your list
Filtered before it reaches you
A finding becomes a decision only if it names one action, stakes at least €50 a week, is under 14 days old and is not a duplicate. A data gap becomes one line on Today, never a decision.
- under €50 a week22
- older than 14 days6
- duplicates4
- data gaps → one line2
7
decisions on Today
Ranked by benefit
What you may decide comes first, then the largest benefit. Weekly, one-off and tied-up money keep their units and are never added together. A decision without a price goes last and says so, instead of showing zero.
≥ 4
control stores behind each proven outcome
You decide, marql checks
Assign a decision and it moves to In progress. When the work is done, the check confirms it and the effect is measured against comparable stores. What it gave lands in Outcomes.
Scored against this location, not a flat rule.
MonTueWedThuFriSatSun
Tuesday, an ordinary day
Tuesdays run about 5% below the week's average here, and this one sits in the middle of its own normal. The tightened flat rule fires on it anyway — and on Wednesday and Thursday.
Saturday, 16% below its own normal
Saturdays run 16% above the average here, so a drop of the same size lands at an ordinary Tuesday's height. A flat −20% rule never reaches it, and tightening one until it does costs three false alarms a week.
A revenue signal fires at 3σ from this location's own normal and at least 15% below it. Saturday is 16% and about 12σ. Illustrative figures.
Everything on Today exists to make the list shorter.
A list nobody can finish is a feed, and a feed gets ignored. The filter, the ranking and the single green button are the product.
Every decision carries its reasoning
Review opens the decision: what marql found, where the money comes from, and how it was computed, with the method, the window and the sources. Then three ways out: assign it, snooze it to next Monday, or dismiss it with a reason that stays on the record. Work from an AI employee opens its own dossier, with Accept, Reject and Talk it through.
You can say no as fast as yes, and the reason is kept.
Needs intervention1 of 7
Review prices and suppliers for low-margin dairy SKUs
Benefit
≈740 € per week
Open for
6 days since Mon
What marql found
Dairy margin fell 4.2 pp in five stores over three weeks. The other seven held.
Written by marql
Where the money comes from
- Dairy sales, the five stores
- 17 600 € / wk
- Margin gap to their own normal
- × 4.2 pp
- Benefit, estimate
- ≈740 € / wk
Your options
If you assign it, it goes to In progress. The amount stays an estimate.
If you snooze it, it leaves this list until next Monday.
If you dismiss it, you pick a reason, and it is recorded with the decision.
How it’s computed
- Method
- Margin against each store’s own 8-week normal
- Window
- 21 days
- Sources
- POS · ERP cost prices · 6 min ago
Decision, In progress, Outcome
An assigned decision waits in In progress with its execution date and the date its effect will be read. The check confirms the work was done, and only then does the measurement window open. No effect counts as measured, never as proven.
Every decision ends in a recorded outcome, including the ones that did nothing.
Decision
assigned 6 Oct
In progress
Done (verified) 8 Oct
Outcome
measured 2 Nov
Receipt · decision #K7Q2MDProven
Move slow stock to two stores
Sell-through of the moved SKUs · 2 stores · 6 Oct – 2 Nov
- Estimate
- ≈380 €
- Measured
- 410 €
- Against
- 4 control stores
How outcomes are measured ↗Share receipt
Move slow stock to two storesProven410 €
Reprice 62 dairy SKUsMeasuring · 16 Nov—
Shift two shifts to Friday peakNo effect0 €
Bundle coffee with pastriesCouldn’t measure—
No-effect results count toward Measured, never toward Proven.
It reaches you
A morning email goes to everyone in the organisation unless a person turns theirs off. The Today badge counts only the decisions you may accept, so a read-only role never sees a number it cannot act on.
You know what waits before you open anything.
marql07:00
7 decisions wait for you · Demo network
Yesterday 26 900 €, +3.4% against the same day a week earlier.
- Review prices and suppliers for low-margin dairy SKUs≈740 € / wk
- Clear dead stock and release tied-up capital≈6 400 € one-off
- Transfer between two stores · AI supply employee≈210 € / wk
On for everyone in your organisation. Turn yours off in Settings › Notifications.
Counts only the decisions you may accept. A read-only role sees no number.
Exactly how much control it takes.
An operator who has been sold automation before wants to know exactly how much control it takes. This is the honest answer.
It proposes.
It does not change your POS or your books. With e:corg, an accepted order or transfer goes into the ERP as a draft for a person to post; with any other ERP you enter it yourself.
The owner or COO decides what each AI employee may do on its own, per type of action, one level at a time.
Detectors are switched on per organisation.
A new network starts with the core set, not the full catalogue.
A supplier order from the AI supply employee is work to check, not a finding, so it skips the €50 floor and can show «Benefit not estimated» next to its purchase cost.
A decision with nothing to measure is tracked until it is done, but it never shows proven money.
The questions that come up in the first demo.
A threshold alert fires on a flat rule — the same number for a kiosk and a flagship. Here each location is scored against its own normal, which the section above draws. Findings are then filtered, ranked by benefit, and most carry a proposed action rather than only a notification. Your own threshold rules can still be added, and they arrive in the same queue.
Data comes in every few minutes. Some checks run after each sync and against the day's pace, the full scans run nightly, and some patterns are weekly. Your AI employees make their pass from 06:00 local time. We publish those intervals rather than calling it real time, because an operator who is told «instant» and sees a four-hour-old number stops trusting everything else on the screen.
Because most of what a detector finds never reaches you. Anything under €50 a week, older than two weeks, a duplicate or a data gap stays off the list. What is left is ranked, only the first decision is open, and the rest wait behind «N more awaiting review».
Roles decide it. An owner and a COO see the network. A store manager sees their own stores and starts on Live; a supply manager starts on Supply. Someone who may not decide sees a decision marked for an authorised person, and no badge.
It became Today. In October 2026 the product folded its inbox into the first screen: the decisions are the queue, and briefs and messages open from the Morning brief row. Old links still work and open the same decision on Today.
Yes. Today, the decisions and the morning email follow each person's own language. The product runs in eight languages, including Romanian, Bulgarian, French, Spanish and Serbian.
It moves to In progress with an owner and an execution date. When the check sees the work done, the measurement window opens, and the result is read against comparable stores: Proven, No effect, Worse or Couldn't measure. A decision that was never carried out is not credited.
Ready to see
the euros it found?
The walkthrough is built around the POS and accounting you already run. Leave your details and pick a time.