Case Study
How B-arm Medicals Brought 9 Branches Onto One Automation Layer
-01%201.svg)
BM Medical Technologies — better known as B-arm Medicals — is a medical products manufacturer that also runs a high-volume e-commerce operation on Amazon, Flipkart, and its own websites. The business runs across 9 branches, each on its own Tally company file, generating more than 10,000 monthly sales transactions and 200–300 purchase invoices that someone has to post by hand every month.
Gokula Krishnan, who runs the business alongside his team, had been looking for a way to automate the back office without ripping out the sales billing format they'd built. In a 31-minute demo on 18 February 2026, that question got answered. Five days later they paid. Six days after that, they were live.
B-arm's volume looks like Distacart's, but the structure is different. Where Distacart is bills-heavy from a centralised AP function, B-arm is sales-heavy from a distributed multi-channel e-commerce engine — Amazon and Flipkart settle payments in tranches that hit weekly, and someone has to reconcile those tranches against purchase orders, returns, commissions, and TDS withholdings.
Three structural problems were eating their finance time:
Nine Tally companies, one human. Each branch ran as its own Tally company file under a single Tally license. Posting a purchase invoice meant opening the right branch's Tally first, then keying in the bill. Cross-branch consolidation meant exporting from each, importing into Excel, and reconciling by hand at month-end.
Bank statements from multiple accounts, weekly. B-arm operates across two operating bank accounts plus credit cards. Statements came in weekly — roughly 7 PDFs a month across the accounts. Each had 300–2,000 transactions inside, many of them Amazon/Flipkart settlement tranches that needed careful tagging to the right sale, the right branch, the right ledger.
No live financial visibility. Gokula wasn't using Tally for any reporting. The reports existed inside Tally — they just required someone to pull them on demand. There was no aggregated view of net profit, burn rate, payables aging, or revenue versus expenses across the 9 branches. Decisions were getting made on lagging information.
Gokula is the kind of buyer who tests before he commits. He asked the questions that mattered to a multi-branch operator with non-standard ledger structures.
Q. "Will I need a separate license for each of my 9 branches, or can one license handle all of them?"
One AI Accountant license covers all branches under a single Tally license. You switch between branch Tally companies inside the AI Accountant interface using an organisation dropdown — same way Tally itself handles multi-company setups. No multiplication of subscription costs by branch count.
Q. "When I upload a purchase bill, can the tool match it to the right branch's Tally company automatically?"
Yes. The tool reads the GSTIN on the bill first — if the GSTIN matches a vendor in a specific branch's Tally master, it routes there. If GSTIN isn't on the bill, fuzzy matching against the vendor name handles routing. Either way, the bill lands in the correct branch's Tally without manual switching.
Q. "Can I customise the expense groupings on the dashboard? I want packing expenses to roll up separately from other expenses, not bundled together."
The default dashboard view groups expenses by Tally's chart of accounts. Custom groupings — like splitting packing expenses into their own bucket — are buildable but require a feature request and timeline estimate. For B-arm, the standard groupings were a starting point; deeper customisation could follow once they were live and had a clearer picture of what mattered.
Q. "Bank statement narrations — when I pay salary to someone named Ram and the bank narration just says 'Ram,' will the tool auto-categorise that to salaries, or will it post to a Ram ledger?"
Initially it would post to whatever the narration suggests. But the tool learns: correct the categorisation manually two or three times, and from then on the same narration pattern routes to the salaries ledger automatically. The training happens by use, not by setup.
Nanditha walked Gokula through every module. The bills upload accepted 500 bills in a single bulk upload, in PDFs, JPGs, or PNGs. Extraction completed at 99% accuracy on properly imaged bills — vendor name, GSTIN, bill number, date, line items with HSN codes, ledger mappings, tax slabs, TDS, all pre-filled. Where TDS wasn't on the bill itself, the reviewer added it in the review screen and approved.
The bank statement module ingested PDF statements and parsed every transaction with date, amount, description, and a ledger suggestion via fuzzy matching against the narration text. For Amazon/Flipkart settlements, the descriptions were standardised enough that after two or three manual corrections the tool would route subsequent settlements correctly without intervention.
The dashboard pulled live from Tally — net profit, burn rate, operating expense, gross profit, revenue versus expenses, cash in / cash out, payables aging by 30/60/90+ days, customer outstanding, and day sales outstanding. All three dashboards (financial, payables, receivables) refresh every minute.
GST reconciliation ran against GSTR-2B fetched directly from the portal using B-arm's GSTIN — no manual 2B download required. Outputs in four buckets: fully matched, AI matched (minor field differences, ITC unaffected), AI probable (flagged in red, ITC at risk), and missing. Exported as Excel for filing prep.
The pricing conversation was where the deal got real. The standard ₹5,000/month plan covers 300 bills and 30 bank statements. B-arm's profile fit exactly:
The ₹5,000 plan handled their volume with margin to spare. They paid on 23 February 2026 and were activated by the next day — vendor masters, ledger mappings, branch-wise Tally connections, GST portal integration, all wired up in under 24 hours.
One honest note. AI Accountant doesn't replace B-arm's sales billing system. Gokula explicitly said he wanted to keep his own billing format for sales — and the right answer was to keep it. The Accounts Receivable module wasn't live for sales invoice creation at the time of the demo. What got automated was the back office: purchase invoices, bank reconciliation, GST matching, and the live dashboard pulling from Tally.
This is the deliberate framing of the product. Sales billing varies wildly by business — every operator has invoice templates, GST treatments, and customer-specific terms baked in. Forcing a sales billing replacement on every customer would slow onboarding from days to months. Instead, AI Accountant integrates with whatever sales system the business already uses (Tally entries, Petpooja for FNB, Amazon/Flipkart settlement files for e-commerce) and automates the work around sales — the parts that are repetitive without being judgment-heavy.
By mid-May 2026, B-arm had been live on AI Accountant for roughly 90 days. The pattern that emerged:
The renewal conversation is currently scheduled. Internal usage signals — sync activity, bills uploaded, GST reconciliation runs — all show the product in active daily use across the operating account team.
B-arm's case isn't about volume saved or hours eliminated, though both happened. It's about whether a fast-growing multi-channel manufacturer can install a single automation layer that scales across branches without multiplying licensing, without ripping out existing sales billing, and without forcing every branch's Tally setup into a new structure.
The answer turned out to be yes — and the onboarding was fast enough (six days from demo to live across all 9 branches) that it didn't disrupt month-end close. That's the proof point that matters for any business in B-arm's shape: e-commerce sales velocity, distributed branch operations, multi-account banking, and a finance function that needs to consolidate it all without growing headcount.
"We are having more than 7 to 10 thousand sales regularly in a month. We would like to automise it as easy as it is."
— Gokula Krishnan, B-arm Medicals (pre-demo brief)
The shape of the value, again, is consistent across customers: the work doesn't disappear, but the manual part does. Reviewers replace data-entry operators. Decisions get made on real-time numbers instead of last-week's spreadsheet. And the finance function stops being the constraint on how fast the business can grow.
