Reconciliation and intelligence for manufacturers, anywhere
Close the month in 4 days, not 12.
Evenfab matches your banks, ERPs, tax filings and factory records into one ledger that is always even, across every entity, country and currency you operate in. AI agents chase customers, fix tax mismatches and prepare the close. People approve. Nothing moves without a signature.
Cash across 14 accounts · 5 countries · 7 entities
$0.0M
Matched on their own0%18,406 lines · 217 for a person
Tax credit found · GST and VAT$0K$96K lapses in 31 days
Collections agent · today$0Kpromises to pay
Refreshing balances across 14 accounts in 5 countries
October close0%day 3 of 6
✦Deccan Coatings paid $14.50 short. That is withholding tax, not a dispute.
0%of bank and ledger lines settle without a person
Day 4statements pack ready, instead of day 12
$0tax credit found stuck in one group's books
0faster collections with the follow-up agent
Illustrative figures from the Evenfab demo environment (a seven-entity chemicals group with plants in India, the US and the Gulf). Your results depend on your data and process.
The problem
Month-end in most mid-size manufacturers is still a spreadsheet marathon.
Bank statements, ERP vouchers, tax-authority downloads and vendor emails are tallied by hand, every month, in every country you operate in, by people who should be analysing the business instead. The books close late, tax credit quietly lapses, and nobody is sure the numbers are even.
- 1Bank lines matched one by oneFourteen accounts, seven entities, three ERPs, four currencies. One payment covering four invoices takes ten minutes to place.
- 2Tax credit that never gets claimedGST credit in India sitting in GSTR-2B but missing from 3B. VAT input credit in the EU and Gulf booked to the wrong period. Every regime has a deadline after which it is gone.
- 3Collections run on memoryReminders go out when someone remembers. Disputes surface at day 90. Statutory payment terms (MSME 45 days in India, late-payment rules in the EU and UK) are crossed unnoticed.
- 4Factory and finance don't agreeBills of materials live in CAD and PLM, yields in the MES, power in meters, and none of it reaches the ledger. Cost per unit is a guess until the audit.
One ledger, every source
Everything that touches money flows into Evenfab.
Banks, ERPs, tax portals, PLCs, CAD and PLM systems, MES and test benches stream in on the left. The AI matches, explains and acts in the middle. What comes out on the right is a ledger that is even, and people who only see what needs a decision.
How Evenfab works
From raw data to a signed-off close, in six stages.
Scroll through the stages. The panel shows what you see in the product at each step.
Stage 01Connect
- 01
Connect
SAP, NetSuite, Dynamics 365 and Tally connectors. Banks by direct corporate API or statement upload, in any country. Tax filings through licensed gateways. Email inbox for invoices and statements.
- 02
Clean up
Seven charts of accounts and four currencies become one group view. "SHARMA PAINTS" in the bank and "Sharma Paints Pvt Ltd" in the ERP become one party, linked by tax ID.
- 03
Reconcile
Exact matches first, then patterns learned from your past months. Differences are explained: withholding tax, bank charges, discounts, exchange rate.
- 04
Agents act
Collections, vendor statements, tax suppliers and accruals. Email, WhatsApp and SMS from approved templates, in the customer's language. Replies are read and understood.
- 05
Approve
Every entry, message, payment run and rule change goes to the right person by type and amount, with the evidence attached.
- 06
Post and report
Approved entries are written back to each ERP. P&L, balance sheet and cash flow per entity, per country and for the group in reporting currency, with plain-words commentary.
What gets reconciled
Fourteen scenarios on day one. Yours get added as the system learns.
Bank
One payment covering many invoices, short payments due to TDS, bank charges with GST, gateway settlements, unknown receipts.
Deccan Coatings paid ₹1,210 short. Exactly 0.1%: TDS under 194Q.
Tax credit · GST, VAT
Input credit in your books but missing from the authority's data. Credit visible but never claimed. Vendors unpaid past statutory limits. Blocked credit.
₹42.6 L sat in GSTR-2B for 5 months. Claim it in the next return.
Receivables and payables
Unagreed discounts, advances before invoice, vendor statements that don't match, the same invoice booked twice, statutory payment terms by country.
Raipur Solvents shows ₹4.22 L due; books say ₹3.64 L. Two invoices not booked, one debit note not theirs.
Intercompany and FX
Both sides of a sale between your own companies matched and eliminated in group statements, across currencies. Exchange gains and losses booked at the bank's actual rate.
Hyderabad → Dubai FZE $98K at 83.10 INR/USD. Matched both sides, FX booked, removed in the group view.
Monthly statements
P&L, balance sheet and cash flow for each entity, each country and the group in reporting currency, with commentary an owner can read without an accountant in the room.
"Power cost is up 11% because Plant 2 ran a third shift." Written by Evenfab, checked by you.
Factory Phase 2
The BOM from your CAD and PLM against what the line actually consumed, test yield against plan, power per unit from the meters. Cost per unit that is reconciled, not estimated.
Lot 4471 used 3.2% more than BOM rev C. $490 you can see on day 1, not at audit.
AI agents, with guardrails
Agents that work like a careful junior, not a black box.
Each agent has a template, a trigger, a list of what it may read, what it may do alone, what needs approval and what is blocked. You test it on last month's data before it goes live. Every run is logged step by step.
- CCollections follow-upChases overdue invoices by email and WhatsApp, logs promises to pay, hands disputes to a person.
- VVendor statement matchRequests statements, matches them to your books, raises differences with the vendor.
- GGST 2B mismatch chaserFinds invoices suppliers haven't filed, asks them to, tracks the 30 November deadline.
- BBank reconciliationMatches every bank line and drafts entries for charges and interest. Daily, no messages.
- AMonth-end accrualsProposes accruals for unbilled power, freight and rent from the last three months' pattern. Drafts only.
Chase everyone over 30 days in Polychem, but ask me before any WhatsApp.
Found 38 customers, ₹3.42 Cr overdue. I checked the banks first: 3 of them paid yesterday and the receipts aren't posted yet, so I've removed them.
Plan: 31 emails I can send now, 12 first-time WhatsApp messages waiting for your approval, 2 handed to people. Kiran Polymers replied that the rate is wrong, so that one is a dispute and goes to Ravi in sales.
Approve the WhatsApp batch. Softer tone for Sharma Paints.
Done. 11 sent, 1 edited. I'll log promises to pay as replies come in and update the cash forecast. Nothing has been posted to Tally.
Allowed aloneRead ledgers and banks, send approved email templates, log promises, escalate to sales.
Needs approvalFirst WhatsApp to a customer, any discount, journal entries, rule changes.
Always blockedPosting to the ERP without sign-off. Initiating any payment, ever.
Hands to a person whenA reply mentions dispute, legal or a manager, or the agent is less than 80% sure what it means.
How we want to proceed
Finance first. Then the factory floor. Then the twin.
The name means tallying: checking that everything is even. We start where the pain is loudest and the data already exists, in any country, and earn the right to go deeper into the plant.
Phase 1 · now
Finance reconciliation and close
- Bank, ERP, GST and intercompany matching across entities
- Collections, vendor and GST agents with approvals
- Monthly statements pack and close checklist
- Design partners: 5 to 8 mid-size manufacturers across India, the US and the Gulf, $15M to $150M revenue
Phase 2 · with design partners
Factory-to-finance reconciliation
- BOM, MES, test-bench and meter data against purchases and sales
- Cost per unit per plant, per line, per batch
- Yield and consumption variances raised like bank exceptions
- Starts with CAD, PLM and MES exports, then PLC and sensor data
Phase 3
Digital twin of the plant's money
- Live view of what each line is earning and consuming
- Machine monitoring tied to cost, not just uptime
- Scenario planning: shifts, raw material prices, power tariffs
- One twin per plant, one ledger per group, wherever the plant is
The digital twin
Every machine on the floor, priced in real time.
The twin maps each line, lot and meter to the money it earns and consumes, starting from the BOM in your CAD and PLM. A yield variance shows up as a cost on the day it happens, in the same ledger as the bank. It is the discipline of CAD geometry and chip verification, applied to a plant's money.
- Cost per unit per lineMaterials from the BOM and MES consumption, power from meters, labour from shifts, allocated the way the line actually ran.
- Variances raised like bank exceptionsBOM vs consumption, test yield vs plan, planned vs actual power. Each one has an owner, a drawing revision and a value.
- Scenarios before you commitAdd a third shift, swap a component in the BOM, absorb a tariff change. See the margin impact per line before it happens.
On BOM and yieldVariance under reviewOver BOM
Built for plants anywhere
One ledger across countries, currencies and tax regimes.
A group with a plant in Hyderabad, a sales office in Houston and a trading entity in Dubai runs three ERPs, a dozen banks and three tax calendars. Evenfab treats that as one reconciliation problem, with local rules applied where they belong.
IndiaGST via GSP · MSME 45-day · TDS
United StatesSales tax · ACH and wires · 1099
EU and UKVAT, VIES, MTD · SEPA · late-payment rules
GulfVAT via FTA and ZATCA · e-invoicing
Southeast AsiaGST and VAT · multi-currency banking
ReportingGroup view in USD, EUR or INR · IFRS or Ind AS
Ground rules
Built for finance teams who have to sign their name.
Evenfab reads your books and banks. It writes back only after approval. Every step is logged with who, or what, did it.
- Bank access is read-only: direct corporate API or statement upload. No account aggregators, no screen scraping.
- No agent can release a payment. Payment files are released inside your bank, with your bank's own approvals.
- Maker-checker on every entry, message batch, payment run and rule change.
- Data stays in the region you choose: India, EU, US or the Gulf. Encrypted at rest. Auditors get a read-only seat with the full trail.
Makeragent or accountant drafts
Checkerfinance manager reviews
ApproverCFO signs above limit
Releasedin your bank, not by us
Blog
Notes on closing faster, claiming what's yours and running a plant on real numbers.
Leaders
Built in Hyderabad, for plants everywhere, by people who have shipped verification systems before.
VNPhoto: use the headshot from finoscale.ai/about on deploy
Vamsikrishna Nukala
Founder & CEO, Evenfab · Co-founder & CEO, Finoscale
IIIT graduate and AI specialist with more than ten years across Microsoft, Synopsys and Siemens. At Finoscale, Vamsi built AI-native credit and risk intelligence used by 100+ enterprise customers in India's financial sector, including KYB, vendor due diligence and reconciliation products.
The throughline of his career is making complex, high-stakes systems provably correct: CAD geometry and PLM at Siemens, chip verification at Synopsys, AI agent systems at Microsoft. Evenfab applies the same idea to a manufacturer's money: every rupee in the bank, the ERP, the GST portal and the plant should tally, and the system should be able to show why.
- EvenfabFounder & CEO. Reconciliation and intelligence layer for manufacturers, finance first.
- FinoscaleCo-founder & CEO. KYB, vendor due diligence, credit intelligence and NBFC lending software; ~2,000 due-diligence reports a month.
- MicrosoftAI agent systems.
- SynopsysRTL and DFT verification for silicon.
- SiemensCAD geometry and PLM.
- EducationIIIT · MBA, Great Lakes Institute of Management.
Advisory board in formation. We are inviting CFOs of large Indian manufacturers and finance leaders from global technology companies to shape Phase 1. If that is you, say hello.
Questions
Things finance teams ask us first.
Does Evenfab replace SAP, NetSuite or Tally?
No. Your ERP stays the system of record. Evenfab reads from it, reconciles against banks, GST and other sources, and writes approved entries back. Your auditors still audit the ERP.
How do bank accounts connect?
Two ways only: a direct corporate banking API where your bank offers one (read-only, IP-whitelisted), or statement upload in PDF, Excel or MT940, including by forwarding to an email address. We do not use account aggregators or screen scraping.
Can an agent move money or post to the books on its own?
No. Agents can read, draft, send approved message templates and prepare payment files. Posting to the ERP needs a person's approval, and payments are released only inside your bank.
What is "stuck" tax credit?
Input credit (GST, VAT) you are entitled to but have not received: invoices in your books that suppliers never reported, credit visible in the authority's data but never claimed, or credit that must be reversed because a vendor went unpaid past the statutory limit. Evenfab finds all three and tracks each regime's deadline, such as Section 16(4) in India.
How long does setup take?
About four weeks: connect sources in week one, learn your books from six months of history in week two, turn on agents in week three, run your first close with Evenfab in week four.
Who is Phase 1 for?
Manufacturers with $15M to $1B revenue, two or more entities or plants, often in more than one country, and a finance team that closes manually today. Chemicals, polymers, auto components, packaging and pharma are where we start, with design partners in India, the US and the Gulf.
Book a demo
See your own month close in 30 minutes.
Bring one bank statement and one month of tax filings from any country. We'll show you what Evenfab would have matched, what it would have found, and what it would have asked you to approve.
Evenfab · built in Hyderabad for manufacturers everywhere. We reply within one working day.