Audit preparation is mostly retrieval — pulling ledgers, matching documents, explaining exceptions. This demo shows TEKAI doing the retrieval conversationally against live SAP Business One data: the questions an auditor asks, answered from the system in minutes, with the exceptions isolated for human explanation.
Not into judgement — into fetching. Trial balances by period. Ledger drill-downs. The document trail behind a journal entry. Party-wise balances for confirmations. The list of transactions above thresholds. Each request is a query someone on your finance team has to build, export and format — and the requests arrive in dozens, multiplied across every auditor question and every follow-up the first answer provokes. The weeks an audit consumes are mostly this: skilled people acting as a slow, manual query interface between the auditor and the system that already holds every answer. Meanwhile the audit clock runs, the fees run with it, and the finance team’s own month-end work waits its turn.
On a demonstration database: auditor-style questions asked in plain language — unusual journal entries this quarter, cash transactions above a threshold, vendor payments without matching invoices — answered directly from the live system. Then the drill-down follow-ups, the way a review conversation actually runs: who posted this entry, show me the documents behind it, which other transactions touch the same party. What surfaces is not a report pack to be assembled, exported and emailed; it is answers, with the document trail attached. The follow-up matters as much as the first answer: an audit review is a conversation, where each answer provokes the next question, and a conversational interface keeps pace with it in a way a queue of report requests never can. Every figure in the demo comes from a demonstration database, not customer data — your own audit runs on whatever your live system holds.
Ninety percent of audit sampling confirms the ordinary — the invoice that matches its PO, the payment that matches its invoice. The valuable output is the exception list: the entries that do not fit the pattern. Pattern-reading across full ledgers is precisely what an AI grounded in the complete dataset does better than a sampled manual pass. Instead of examining a fraction of transactions and hoping the sample catches the problems, the sweep reads everything and returns only what deserves a professional’s attention. The human hours move from confirming the ordinary to explaining the exceptional — which is where they were always supposed to go. Full coverage also changes the character of the assurance: a sample says “we found no problems in what we looked at”; a sweep of the whole ledger says “we looked at everything.”
The same questions that serve the statutory audit serve a monthly self-audit. Run the exception sweep before the auditors do: unusual entries, threshold breaches, unmatched payments, caught while they are one month old instead of eleven. Issues get fixed while the context is fresh, and year-end becomes confirmation instead of archaeology — the audit team walks in to find a system whose exceptions have already been found, investigated and documented. The check that once justified an annual project now costs no more than asking the question — and that change in cost is what changes the frequency.
SAP Business One (SQL or HANA) · TEKAI at ₹5,000/company/month India, $150 global · your own AI subscription (Claude, ChatGPT or Perplexity) · permission-scoped, read-path access for audit-style queries, nothing trained on your data (/tekai/security/).
Access is yours to grant — a scoped, read-only audit user is the natural pattern, so the audit team asks its own questions without touching anything else.
No — it replaces the retrieval labour. Opinions, assurance and judgement remain professional work; the AI just stops that work waiting on exports.
Whatever the permission scope allows — ledgers, documents, masters — and nothing more.
Yes — the monthly exception sweep is the same capability pointed inward. The fastest way to see it on your own data is the five-minute Fit Test.
New to the category? Start with the overview: AI for SAP Business One, explained.