This demo shows Claude building finished deliverables — Excel workbooks, PowerPoint decks, Word documents — directly from live SAP Business One data through TEKAI. Not exports you still have to reformat: the formatted deliverable itself, produced from a plain-language request.
Board packs live in Excel, PowerPoint and Word — so that is where the ERP data has to end up. Usually by copy-paste, late at night. This is the demo where that stops.
The data was never the slow part; the deck was. Numbers leave SAP Business One clean in seconds and then eat an hour of formatting: paste into the workbook, fix the columns, rebuild the chart, restate the summary slide, adjust the commentary. The monthly review pack, the bank presentation, the board summary — each one rebuilt by hand every cycle, by someone whose actual job is analysis, not layout.
And because the formatting hour is manual, it gets scheduled — which means the numbers inside are already a day old when the pack is presented. The slow last mile doesn’t just cost time; it costs freshness. The pack that matters most — the one going to the board or the bank — is exactly the one that gets rebuilt most carefully, and therefore most slowly.
On a demonstration database, requests like these produce finished files: a sales-review deck for the quarter, a receivables workbook broken out by aging bucket, a one-page performance summary. The output is actual .xlsx, .pptx and .docx files — populated from live data pulled through TEKAI at ask-time, structured the way the request asked: tables where tables belong, charts where charts belong, commentary written from the figures.
The request is the specification. “A quarterly sales review deck, one slide per region, with a summary of movements” is enough — and refinements are just more conversation, not another round of manual editing. Ask for the receivables workbook with one tab per aging bucket and a summary sheet on top, and that is the shape of the file that arrives.
The commentary is part of the build, not an afterthought: the document arrives with the figures explained in prose written from the data it contains, ready for a human to review and sharpen rather than to draft from a blank page.
Every figure in the video comes from a demonstration database, not customer data.
A CSV export is homework; a formatted deliverable is done. That is the whole distinction. An export means the hour of layout work still stands between the data and the meeting — the deliverable means it doesn’t.
The difference compounds monthly. The review pack that took an afternoon becomes a request that takes a sentence — twelve afternoons a year returned to the finance team from one recurring document, before counting the ad-hoc requests that currently get declined because nobody has the afternoon. And it’s never stale, because it’s built at ask-time from live data: run the same request an hour before the meeting and the numbers are that hour’s numbers.
SAP Business One (SQL or HANA) · TEKAI at ₹5,000/company/month India, $150 global · a Claude subscription — this workflow is Claude-specific, because file creation runs in Claude’s environment · permission-scoped access, nothing trained on your data (/tekai/security/).
Excel, PowerPoint and Word in the demo — built as real .xlsx, .pptx and .docx files, not screenshots.
Structure and content follow the request; house-template fidelity is a matter of instruction detail — the more precisely you describe the format, the closer it lands.
Yes — pulled through TEKAI from SAP Business One at the moment of the request, so the deliverable is never built from a stale export.
Recurring deliverables ride the same delivery patterns as TEKAI reports.
File creation is a Claude capability; other assistants connected via TEKAI answer questions, but this workflow builds documents.
New to the category? Start with the overview: AI for SAP Business One, explained.