How to connect Allegro to ChatGPT and Claude: stock, sales and orders

How to connect Allegro to ChatGPT or Claude

Friday, two hours before the courier comes, and your best-selling mug set might not last the weekend. If Allegro is linked to ChatGPT or Claude, the chat can tell you how much is left, how fast each offer sells and which paid orders still need to go out. Neither ChatGPT nor Claude had an official Allegro connector in its catalog when we checked on 23 October 2026. APIUniCon is how you add one yourself.

1. Add APIUniCon to ChatGPT or Claude; the setup guide has the buttons. One setup covers every service you connect later.
2. Write in the chat: “Connect my Allegro account”. If the AI asks where Allegro describes its methods, give it the Allegro developer portal.
3. Approve access on Allegro. This step is different from most services: there is no key to create. Allegro’s authorization guide (in Polish) says an outside app may not ask you for your own Client ID and Client Secret. It has to use its own and ask for your consent on allegro.pl. So the link the AI sends opens our page, and our page takes you to Allegro’s sign-in. Log in as usual, read the list of permissions on Allegro’s consent screen and agree.
4. That’s it. The AI makes a test request and says what it can read.

Your password stays with Allegro. What APIUniCon receives is a token that Allegro issues after your consent, and we store it encrypted.

What to ask about your Allegro offers and orders

Write the way you’d ask whoever runs the shop with you:

• Which offers will run out within 7 days at the pace of the last 30 days?
• How many days of cover do my 20 best-selling offers have?
• Which paid orders have no tracking number yet, and which must be dispatched today?
• Which offers had more than 300 visits in the last 30 days but sold fewer than 5 units?
• How many units of SKU KM-204 did we sell in the last 30 days, and at what average price?
• Which offers have the most people watching them but no sales in the last 30 days?
• What did buyers return this month, and which return reasons come up most?
• How does this week’s sales value compare with last week’s?

Cases before the courier comes

Everything below happens in a made-up homeware shop. Your answers will show your own offers.

Will it last until Monday’s delivery?

The supplier delivers on Monday. Before the weekend the owner asks: “Which offers run out before Monday at the pace of the last 30 days?” For every offer Allegro reports two things the AI needs: the stock available now and the units sold in the last 30 days. Divide the second by 30 and you have the daily pace.

Two offers won’t make it to Monday. The 6-piece mug set has 7 left and sells about 3 a day (90 in the last 30 days), so it runs out on Sunday. The linen tablecloth has 4 left at 2 a day (60 in 30 days) and will be gone by Sunday as well. The bamboo board looks low at 14 units, but at 2 a day that is a week of cover.

A 30-day average can lag behind a sudden rush. If an offer took off this week, ask the AI to count the pace from the last 7 days of orders instead.

Paid, but not shipped

At two o’clock the question is about the parcels: “Which paid orders have no tracking number yet, and which must be dispatched today?” The AI takes the orders with completed payment where no item has a tracking number. Allegro gives each one a latest dispatch time, so the answer can be sorted by urgency.

11 paid orders have no tracking number. 4 must be dispatched today: two to parcel lockers and two by courier. 5 are due tomorrow. 2 were due yesterday, both for the linen tablecloth, so those buyers are already waiting longer than promised.

When the parcels are packed, hand the AI the tracking numbers and it adds them to the orders. More on that below.

Visitors who don’t buy

Some offers get plenty of traffic and almost no orders. Allegro counts the unique users who viewed each offer in the past 30 days, and the AI compares that with the units sold. Asked “Which offers had more than 300 visits in the last 30 days but sold fewer than 5?”, it might say:

Two offers. The oak serving tray: 640 visits, 2 sold, 31 people watching it at 189 zł. The glass jar set: 410 visits, 4 sold. For comparison, the mug set sold 90 from 1,150 visits. People find the tray and leave, so check its price and main photo before anything else.

The AI can’t tell you why a visitor left.

What the AI can do in Allegro

You can also ask for changes. For each one the AI shows the offer or order with its before-and-after values, and nothing reaches Allegro until you confirm.

• “Set the mug set’s stock to 40, the delivery has arrived.” You see the offer with its current and new quantity first.
• “Lower the oak tray to 169 zł.” The old and the new price come up for you to confirm.
• “Add these four tracking numbers to today’s orders.” The AI matches each number to an order, shows you the pairs and adds them once you agree.
• “End the offers for the enamel pots, we’ve stopped selling them.” The list of offers comes first, and they end only after your go-ahead.
• “Reply to the buyer who asked about delivery to Kraków.” The AI drafts the message. You can edit it, and only the text you approve is sent.

Allegro applies price and stock changes as a batch and reports on each offer. The AI reads that report and tells you if Allegro refused a change, so you don’t have to open the account to find out.