The platform
Every Indian and global dataset, for specialised workflows
Validated equities, economic and market data, joined and kept current — and read for what is new on the names you follow. That incremental layer is what the platform works from: agentic orchestrators run deep research on a schedule or answer a quick follow-up against it, and every claim is traceable, so you decide with confidence.
01 · Sources of information
Everything that arrives, checked before it's used
The accuracy of the research is decided here, before a single question is asked. All of it is joined: a question can move from something said on a call, to a line in the accounts, to an employment series, without leaving the answer or losing its source.
Live earnings calls
Listened to as they happen and written up within minutes, with every speaker identified. Coverage runs well past the index, into companies where research coverage doesn't reach.
Announcements & filings
Exchange disclosures taken in directly — pledges, shareholding changes, related-party transactions, auditor changes. Structured, free, and mostly unread.
Research notes
What the desk already receives, arriving through its own channels — compressed to the facts once, then opened up wherever the houses disagree.
Company accounts
Standardised statements, segment data and the per-unit measures that matter in each industry, going back years.
Economic & employment data
Rates, commodities and capital-expenditure series, alongside employment figures read as a company, sector and hiring signal.
Fund holdings & liquidity
Disclosed holdings across every fund house and scheme, and what a position can realistically be built or exited at.
Global markets
International indices, commodities, currencies and global peers, for the companies whose drivers sit outside India.
Your private data
Internal notes, past analysis and models, brought in when you're ready — the layer no one running the same query elsewhere can reproduce.
02 · Research
From a quick follow-up to deep research
Quick questions
Conversational enquiry at speed — anchored to something you just read, a name from a search, or nothing at all. Cheap to start and free to abandon, which is what makes people actually explore rather than leaving to do it elsewhere.
Agentic research
A commissioned sweep across every dataset, every call and outside sources, with agents running and every claim sourced. Start it cold or escalate a conversation into it. Schedule it by sector — and have a scheduled question arrive by email as your morning brief.
Search — by number and by language
Screening on fundamentals is table stakes. Searching on what was actually said is not: find every company whose management changed how it talks about pricing power, or where a capacity plan was mentioned but never quantified.
What to watch, and what to ask
Set the conditions that matter to you and they're applied wherever you scope them. Separately, the system raises what it finds genuinely novel and suggests the question to put next — including while a call is still running.
Where the research disagrees
Fifty houses cover a name and ninety per cent of what they write is the same. We keep the facts once, then surface only what differs — the dissenting read, the risk one house names, the forecast that breaks from the pack.
Working tools
Run an analysis across a company's documents, review several companies at once against one agenda, pull tables out of a PDF, and reuse your own report formats.
03 · Coverage
Three perspectives, side by side
What a company says about itself, what the analysts covering it say, and what you concluded — held together on one page, so the story you decide on is the one that survives all three.
What management says
From the annual report and every call the company has held. Including what it guided to in earlier quarters against what it went on to deliver.
What the sell side says
Fifty houses, ninety per cent identical — kept once as facts, then opened only where they differ. And how each house's earlier forecasts actually turned out.
Your own view
Timestamped and kept, never overwritten, so you can look back at how your thesis stood at any point — and compare it against how the other two moved.
Conviction is what's left when the three agree. Where they don't, that's the question worth asking.
Company accounts · what you expected before the result and how it differed · the call and what was said · research received on the name · management meetings and channel checks · and the sector and its competitors read relatively, so you can tell a company problem from a sector one.