Next-gen Orchestrated Virtual Analysts
An AI analyst trained on the finance canon, the models and the workflows your desk runs on. It connects to the tools you already use, takes a brief in plain English, and learns your firm as it goes.
Why this is not a chatbot
Every Indian company that has ever filed, structured into fields you can query — then reasoned over. No general assistant has this, because nobody else has built it.
of 3.0M ever registered in India — 2.8M of them structured by NOVA, live or struck off.
Terminals cover the listed sliver — about 5,300 names. Your targets, suppliers, borrowers and comparables sit in the other 99.7% — filed, scanned, and until now unusable.
One prompt, a hundred companies
Because it is structured, one question can span the universe: private cement makers above ₹300cr, leverage under 3x, no charge filed in two years — ranked, with the arithmetic behind every filter.
How it works
The register is the floor, not the product. What makes it yours is the same five things you teach a new joiner — taught once.
A new analyst gets logins on day one. NOVA gets the same set — and reads only what the person asking may read.
You walk someone through the weekly pack once and expect it back next Monday. Hit record, work your way, and NOVA keeps the sequence — and the reasoning behind each step.
Which WACC, which convention, what counts as normalised. Teach it once; it applies them, and marks them as assumptions rather than facts.
The formatting fight is the last hour of every job. NOVA learns your formats and section order from work you have already signed.
Ask a good analyst and you get the filing, the note, the page. Ask NOVA and you get the same — plus the arithmetic.
AOC-4 FY25 · NOTE 14 · PAGE 43
Read from the filed accounts. The cash flow statement in the same filing says ₹2.54cr less — flagged, not averaged.
What a tool cannot do
A learning layer runs inside your perimeter, watching what you accept and correct — and what one desk teaches it, the whole firm inherits. Six months in, it opens the quarter before you ask.
Pull Q3 into the DCF and update the drivers.
Refresh Q3 in Apex_DCF as usual?
Q3 is in, drivers updated, three cells flagged.
And it is one analyst, not thirty
When one person saves the way they run a covenant pack, the next person who asks for that work gets it done that way. Junior turnover stops costing you the method.
SAVED AS A SKILL
Weekly credit pack — R. Mehta
CREDIT
PE
ADVISORY
Same method, whoever asks.
Before it leaves
Names, counterparties and figures are replaced by tokens inside your perimeter. The model reasons over the shape of the question; the answer is re-hydrated on your side.
IN YOUR PERIMETER
Model the Apex ComponentsENTITY_1 buyout at ₹6,340crVALUE_1.
Trust
The first question is never whether it works. It is where the data sits, who can see it, and what leaves.
Your perimeter · on-prem or VPC
01Zero retention. Your deal data never trains our models.
02Encrypted in flight and at rest, keys held per tenant.
03Deal-team walls hold. NOVA sees only what you see.
04Runs inside your perimeter — single tenant by default.
05Rolled out by ex-bankers, not a support ticket.
Where it runs
A workspace that hands back finished artefacts, a panel beside your sheet, and an API behind your own tools.
Bring a company you know well and a template you actually use. We will show you the first pass, and where NOVA disagrees with the documents you were given.
Forty-five minutes, your data stays yours, no procurement conversation required.
* Required
Thanks — that’s with us. We’ll come back within a day or two with two slots and a short list of what to bring.
Curated by people from
We're rolling out NOVA carefully — quietly piloting with PE, IB and audit teams who care about getting it right. Drop your details and we'll be in touch within a day or two.
* Required
We'll come back within a day or two with two slots and a short list of what to bring. Bring a company you know well and a template you actually use.