home / use cases / data, spreadsheets and reporting / kpi-alerts

KPI alerts with context, not noise

An Aivell checks your key numbers against your own database every hour, compares them with the usual pace, and speaks up only when a threshold is crossed — with context: what moved, since when, and how unusual it is. No dashboard to open, no daily noise, nothing leaving your network.

Watch the demo

starts when: Every hour, the schedule checks the numbers

The problem

Metric-watching fails in two directions. Either there’s a dashboard that nobody opens — the refund spike sat there for nine days, plotted beautifully, unseen — or there’s an alerting rule that cried wolf twice a day until everyone filtered it to a folder. Both end the same way: you learn about the problem from a customer, or from the month-end numbers.

The middle ground takes judgment. “Tuesday mornings are always slow” is knowledge a threshold doesn’t have; “under pace for a Tuesday” is the alert you actually want. Encoding that in a monitoring tool is a project. Saying it to a colleague takes one sentence.

How your Aivell does it

  1. The schedule runs the checks. Every hour through the working day, the Aivell queries the numbers the chore names — orders, revenue, refunds — over a read-only connection.
  2. The baseline is the same weekday, not yesterday. Today’s pace is compared with the last four of the same weekday, so Monday isn’t judged by Saturday’s standards.
  3. Thresholds are your sentences. “20% under pace, tell Paolo.” “Any single order over €10,000, flag it.” “Refunds over 3%, name the products.” Written in plain language, changed in seconds.
  4. Alerts arrive with the story attached. Not “metric breached” — but the number, the baseline, when the drift started, and what’s driving it. You decide in one read whether to act.
  5. Silence is respected. Normal days produce nothing. A repeated condition doesn’t re-alert for 24 hours. Friday brings a one-paragraph recap so quiet weeks are confirmed, not assumed.

The guardrails keep it honest: the database connection cannot write, and the mail cannot leave your domain. Your trading numbers are computed onboard and stay behind your firewall — no metrics platform holds a copy. A colleague who watches your numbers all day, inside your walls, and only taps your shoulder when it matters.

The chore, as you'd write it.

No flowcharts, no code — a chore is just a message to your Aivell, in your own words. It shapes it into a solid, guarded procedure and follows it to the letter. This one:

Giulia Aivell · watch the numbers ● active

every hour between 8 and 20, check the day's order count and revenue

compare with the same weekday over the past four weeks

revenue tracking 20% under the usual pace? email Paolo with the numbers

a single order above €10,000 — flag it for a manual check, quietly

refund rate above 3% on any product — alert, with the top three offenders

stay silent when everything is normal; no daily all-fine mail

if the same alert fired in the last 24 hours, don't repeat it

friday evening, send a one-paragraph recap of the week's alerts

You
Got it — I'll start right away, and check with you whenever something needs a human.
Giulia Aivell

The tools it uses.

Connected with single-click passkeys — no copied tokens, no OAuth hell.

Schedule runs the checks through the day
Database reads the metrics, read-only
Email delivers alerts and the weekly recap

Sensible guardrails.

Every tool can be limited to exactly what this chore needs. For this one, you might set:

Database Read-only, and only on the reporting views. The order and customer tables proper are out of reach.
Email May write only to Paolo and the operations list. Nothing leaves the company domain.

Questions, answered.

How is this different from the alerts in my dashboard tool? +

Two ways. The rules are sentences — "20% under the usual pace for this weekday" — not formula builders. And the alert arrives with context already attached — what moved, since when, against what baseline — so you don't open a dashboard to find out why it fired.

Will it flood my inbox? +

Silence is the default. The chore says when to speak — thresholds, deduplication within 24 hours, a weekly recap — and you tune those lines the way you'd brief a colleague who pinged you too often.

Our margins and refund rates are nobody's business. Where does this run? +

Entirely on the box. It queries your database across your own network and reasons onboard. No metrics platform, no analytics vendor, no outside copy of your trading numbers.

Which databases can it watch? +

Postgres, MySQL, Oracle, SQL Server — anything reachable with a standard driver — plus Google Sheets, if that's where a number lives.

Related.

What is an Aivell?

An Aivell is an on-premise AI colleague: a small box that plugs into your network, runs its own AI onboard and takes the repetitive work off your desk. Prompts, documents, data — nothing ever leaves your office. Unplug the internet: it keeps working.

You describe each task as a chore, in plain language. Your Aivell turns it into a solid, guarded procedure and runs it in the background — with guardrails on every tool, a complete audit trail, and approvals in your hand when you want them. One-time setup, fixed monthly fee. No tokens, no overages.

Watch the demo

onboard inference · air-gapped · fixed monthly fee