Most audit software starts from an export: someone pulls a trial balance or a day book out of Tally into Excel, and everything downstream trusts that nothing was dropped on the way. Ledger Analysis starts one step earlier, with the books themselves — read whole, and the read proved.
The books read whole
Audcrix reads a client's ledgers and vouchers directly from Tally over a paired outbound channel. No inbound port is opened on the Tally machine, which is the question every client's IT person asks first.
- The whole population, including hundreds of thousands of vouchers, read resumably so a dropped connection picks up where it stopped.
- Incremental synchronisation, so a second read fetches only what changed.
- Prior-year closings pulled automatically, so openings can be tested rather than assumed.
- A wrong-company guard that stops a read before anything loads if the open Tally company is not the client you meant.
- A Day Book XML and trial-balance upload fallback for the day a live connection is not possible.
The read proved
A synchronisation is not trusted because it finished. It is reconciled against Tally's own trial balance and profit and loss, and a mismatch blocks finalisation instead of flowing quietly into every figure built on it.
Every ledger classified by what it is
Each ledger is placed by its own group chain in the books — never by a keyword in its name, because a ledger called "Bank Charges Payable" is not a bank. From there:
- Abnormal balances stay on their natural side, so a debit sitting in a creditor is shown as the exception it is.
- Debits net against credits, never absolute values, so a ledger that swung both ways during the year is not overstated.
- Movement, balance and behaviour are analysed ledger by ledger.
- A trial-balance tie-out, including the implied closing stock where inventory is integrated.
- Combined vouchers are split per party leg, so one entry paying several parties is attributed to each of them.
The base everything else reads from
Every other module — scrutiny, accounting checks, TDS, disallowances, GST — works from this one classified dataset. Every headline number on every screen opens into the ledgers, the vouchers and the group chain that produced it, so a figure is never something you have to take on trust.
Questions CAs ask
Do we need to open a port on the client's Tally machine?
No. Audcrix reads over a paired outbound channel, so no inbound port is opened on the Tally machine.
How do we know nothing was dropped during the read?
Each synchronisation is reconciled against Tally's own trial balance and profit and loss, and a mismatch blocks finalisation rather than flowing into the figures.
Can it work without a live Tally connection?
Yes. A Day Book XML and trial-balance upload is available as a fallback. Zoho Books is also supported as a books source.
How are ledgers classified?
By each ledger's own group chain in the books, never by a keyword in its name, with abnormal balances kept on their natural side.
Further reading
Related modules
- Detailed Scrutiny — Transaction-level voucher scrutiny across the whole Tally population — exception flags, period read from the narration, and prior-period items confirmed by the CA.
- Accounting Checks — Completeness, cut-off, accrual and classification checks run over the whole Tally books — every verdict showing its basis, every finding exportable as a working paper.
- GST Reco — Reconcile Tally books with GSTR-1, 2A, 2B and 3B period by period — register-led outward supply, every leg classified by its own tax tag, and a colour-coded matrix.
