📄 Download the twelve prompts: The Owner’s 12 Hardest Questions for SAP Business One — ready to paste into the AI you already use. Jump to the download →
Every business owner carries the same twelve questions around. Where is my cash? Which customer is about to hurt me? Why did profit fall when sales went up? Is my month-end actually clean? None of them are hard to ask. They are hard to answer, because in most companies the answer means a request to the accounts team, a week of reports, a spreadsheet nobody fully trusts, and a number that is already stale by the time it reaches the owner’s desk. So the questions go unasked, and the owner runs the business on instinct and the bank balance.
In the video above, I ask all twelve of a live SAP Business One system, in plain English, one after the other, in a single sitting. No reports were built. No screens were learned. The AI (ChatGPT, Claude, Gemini or Copilot, whichever you already use) is connected to SAP Business One through TEKAI, and every answer comes back with the SAP document numbers you can open and check. Where an action is needed, the AI drafts the email, the quotation or the purchase request, and a person approves it before anything moves.
The company in the video is SAP’s demo company, OEC Computers India, not a customer’s system. Everything below is what the AI found there.
Find me ₹50 lakh in 30 days. The AI worked through open receivables by due date, customer payment history and disputed items, and came back with ₹61 lakh of cash reachable within 30 days, ahead of the ₹50 lakh asked for. It flagged three invoices to verify before anyone calls the customer, because chasing a disputed invoice costs more than the money it brings in. It then drafted the collection emails, ranked by amount and likelihood, for a human to send.
Which customer will hurt us next? This is the question a credit-control report never quite answers, because the report looks at overdue balances rather than direction of travel. The AI found a customer whose payment days had moved from 70 to 87 to “not paying at all”, while its order value grew from ₹0.6 lakh to ₹92 lakh over the same period. Growing orders and lengthening payment is exactly the pattern that turns into a bad debt, and it is invisible in an ageing report because each individual invoice still looks recent.
What if the rupee drops 5%? The AI pulled every open foreign-currency receivable and payable, applied the movement, and gave the net exposure with the documents behind it: roughly ₹1.7 lakh at risk on a 5% move. Small on this demo company; on an importer it is the number that decides whether to hedge.
Sales went up, so why did profit go down? Gross margin was down 2.9 points. The AI split the fall between price and cost, and the answer was uncomfortable: all of it was given away at the point of sale through discounts and price-list exceptions, and none of it came from supplier cost. That is a sales-discipline problem dressed up as a market problem, and it is a very different conversation to have with the team.
A supplier raised prices 12%. The AI identified every item bought from that supplier, the finished goods that use them, the margin on each after the increase, and the alternative vendors already in the system with a purchase history. Then it drafted the purchase enquiry to the alternatives.
Should I accept this 500-unit order? Before saying yes, the AI checked stock on hand, open purchase orders, the customer’s credit position and margin at the quoted price, and gave a recommendation with the reasons. The same check by hand takes four screens and a phone call to the warehouse.
Plan my festival stock. Using last year’s festival-period sales, current stock and what is already on order, the AI produced a buy list by item with quantities, and drafted the purchase requests for approval.
Which quotations did we lose, and why? ₹1.52 crore of quotations had expired without becoming orders. The AI grouped them by reason, and ₹60 lakh of that was lost to our own price list: quotations issued above the price the same customer had been given before. That is not a lost-to-competitor problem; it is a fix in the master data.
Which customers stopped buying? The AI listed customers with regular past orders and nothing recent, with what each was worth, and drafted the win-back emails, worth about ₹30 lakh of annual revenue on this demo company.
Is my month-end clean? No posting period had been locked since 2008, and a delivery was about to be invoiced twice. Both are the kind of thing an auditor finds in March and an owner would rather find in the first week of the month.
Show me anything unusual. With no direction at all, the AI scanned the last 30 days for outliers: unusual discounts, documents posted outside working hours, quantities out of pattern, journal entries with no reference. This is the question owners most want to ask and least know how to phrase for a report writer.
Prepare my board pack. Finally, the AI assembled the findings from the previous eleven questions into a board deck, with the numbers, the document references and the recommended actions. You can open the deck it built here: board deck.
None of these answers needed a query, a Crystal report or a dashboard. The AI reads the SAP Business One system directly through TEKAI, using the Model Context Protocol, and it respects the SAP Business One authorisations of the user who asked. If you cannot see a company, a warehouse or a price list in SAP, the AI cannot either. If you cannot post a document, neither can it. Actions are drafted, not executed, until a person approves them.
TEKAI works on the SAP Business One version you already run, SQL or HANA, with no upgrade and no add-on installation on the client. Enabling it takes under an hour, and you bring your own AI subscription: India ₹5,000 per company per month, global USD 150.
All twelve prompts are in a free PDF, written so you can paste them straight into your AI as they are or change the numbers to your own. Fill in the form below and the link comes to your work email. Business domains only — we don’t accept Gmail, Outlook or other personal addresses.
Do I need to build reports for the AI to answer these?
No. The AI reads the SAP Business One system directly through TEKAI; no reports, queries or dashboards are built.
Which AI does this work with?
ChatGPT, Claude, Gemini or Copilot, on your own subscription.
Can the AI change data in SAP Business One?
Only within the user’s SAP Business One authorisations, and only after a person approves the drafted action.
Does this need an SAP Business One upgrade?
No. TEKAI works on your existing version, SQL or HANA.
If you would rather see it on your own questions, book a live demo and bring the three questions you have never been able to get answered.
Further reading: AI in SAP Business One: the complete guide · TEKAI · tekai.app
TEKROI Private Limited is an official SAP Business One partner (SAP PartnerEdge — Sell & Service). All data shown is SAP’s demo company. SAP and SAP Business One are trademarks of SAP SE.
12 prompts, ready to paste into ChatGPT, Claude, Gemini or Copilot connected to your SAP Business One system — cash, credit risk, FX, margin, supplier pricing, order acceptance, festival stock, lost quotations, lapsed customers, month-end, anomalies and the board pack. Each prompt comes with what to expect back and which SAP documents the answer should cite.
Your download is ready — we have emailed you the same link.
Download the PDF →This link is valid for 15 minutes.