Operations
Stock alerts that arrive before you are out
Polaris does not hold your inventory. A worker reads the table you already keep, compares it to what is selling, and raises a task when cover gets short.
What the worker does
An AI worker in Polaris tracks inventory by reading the stock table you keep in Supabase alongside sales in Stripe, computing days of cover per item, and raising a task with an owner whenever cover falls below the threshold you set. It posts the stock report as a comment and alerts in Slack. Polaris is not an inventory system and does not write to your stock records.
- Connections
- Supabase · Stripe · Slack
- Computes
- Days of cover per item
- Writes to your stock records
- Never. Read only
Most small operations already have the data and no alarm on it
A stock table exists. Sales exist. What does not exist is anything that looks at both on a Tuesday morning and notices that the item selling four times faster than last month has eleven days of cover and a six-week lead time.
That gap is why reorder decisions get made in a panic, at a worse price, with expedited shipping.
A worker closes it by doing a boring calculation on a schedule and turning the result into an owned task rather than a notification. A notification gets dismissed. A task with a name and a date on it in the Focus lane does not.
What the worker computes and reports
Days of cover per item
Current stock divided by the recent sales rate, using the window you specify rather than a lifetime average that hides a trend.
Items where the rate changed sharply
A doubling in velocity matters more than an absolute level, and it is the signal a static reorder point misses.
Cover measured against lead time
Thirty days of cover is comfortable at a one-week lead time and an emergency at a ten-week one. The lead time comes from your data, not from an assumption.
Items with no movement
The other half of the problem, and the half nobody asks about.
Where each number comes from
| Input | Source | What it must contain |
|---|---|---|
| Current stock | Your Supabase table | An item identifier, a quantity, a timestamp |
| Sales rate | Stripe, or your own sales table in Supabase | Enough history to cover the window you chose |
| Lead time | Your own data or the vendor register | Days, per item or per supplier |
| Threshold | You set it in the acceptance criteria | A number of days, per item class |
| Owner | You set it | A named person who receives the reorder task |
What Polaris is and is not here
What it does
- Reads your existing stock and sales data
- Does the arithmetic on a schedule
- Raises an owned, dated task when a threshold is crossed
- Posts a readable stock report as a delivery comment
- Alerts in Slack where the team already is
What it is not
- An inventory management system
- A warehouse or fulfilment tool
- A writer to your stock records
- A purchase-order system
- A replacement for counting what is on the shelf
Questions people ask
+Do we need Supabase to use this?
You need your stock data somewhere the worker can read it, and Supabase is the database connection in the fixed catalog. Teams already keeping stock in a spreadsheet in Google Drive can point the worker there instead, with the same caveat about how fresh the numbers are.
+Can it place the reorder?
No. It drafts the purchase enquiry and raises the task. Committing company money to a supplier is a human action, and the same rule applies here as everywhere else in Polaris: agents deliver, humans close.
+How often should it run?
Daily for fast-moving items, weekly for everything else. Billing is by delivered human-equivalent hour, so a short daily arithmetic job is inexpensive, and each run is logged with its own line you can check.
+What if we have multiple locations?
Then the stock table needs a location column and the acceptance criteria need to say whether cover is computed per location or in aggregate. Getting that decision written down explicitly is worth more than any feature.
Related
Operations work with an AI worker doing the chasing
Ops is the function that holds the seams together. Most of that work is asking people things and writing down what they said.
Procurement comparisons built from quotes, not from vendor websites
A worker researches the field, collects what the vendors actually told you, and lists the questions you have not asked yet. You choose.
The null rate tripled in March and the board slide was already wrong
Row counts, null rates, orphans and freshness, checked every morning. Silence means clean.
Budget reports written for the people who did the spending
Plan against actuals is easy to produce and hard to read. A worker writes the variance up in sentences a department head will actually act on.
Hire an AI data analyst
It settles what active user means, writes the query that matches the definition, and turns the numbers you hand it into a paragraph an executive can read.
Connect Supabase to Polaris
Supabase is the connection where a careless key choice does the most damage, and the only one Polaris cannot yet check for you.
Connect Stripe to Polaris
This is the one connection where the key you choose matters more than anything on this page.
Polaris for ecommerce brands
You staff for peak and pay for it in February, or you staff for February and suffer in November.