Platform · Product data

Your products are already written down. Somewhere.

Your suppliers send price lists in forty different layouts. Taught once per supplier, read into structured product data with images, checked by a person.

The real blocker

They are in forty price lists from forty suppliers.

Dense article tables

Hundreds of rows, printed small, with the code and the rate three columns apart.

Finish-and-size matrices

One product across a grid, where the price lives at the crossing of two axes.

Card-style catalogues

A picture, a name and a number per tile — laid out for the eye, not for a machine.

Excel with merged cells

And headings that apply only until the next heading appears, without saying so.

This is why most businesses in these trades never get a product database. Not the cost and not the software — the weeks of retyping that nobody has, and the errors nobody catches. A price typed one digit wrong does not announce itself.

It sits there until a customer is quoted from it.

Show it one page

1

You show it one page

this is the item code, this is the price column, this is the size
2

It reads the rest

recognising pages laid out the same way, however many hundred there are
3

The profile is saved

what it learned about that supplier, kept against his name
4

The next edition starts done

largely finished before anyone opens it — and §5 is why that matters

The same file gives the same answer, every time

This matters more than it sounds, because the obvious alternative is worse than doing nothing.

A general AI tool
  • Gives you an answer — it will look right, and it will be right most of the time
  • And then, on page ninety-four — a price confidently attached to the wrong product, perfectly plausibly
  • Rubbish you can see is a nuisance — rubbish you cannot see is expensive
A deterministic engine
  • The same document, the same result — every time, from rules it was taught
  • Not from a judgement it made — which is the entire difference
  • Rules you can inspect — taught from your supplier’s own page, not inferred from somewhere else

Where it cannot read something, it says so instead of inventing it.

Every row is clean or flagged

This code breaks the pattern

It does not match what was learned from your supplier’s own list. Raised, not guessed at.

This price looks wrong

Out of range for the rest of the range it sits in. A digit is easy to mistype and impossible to spot.

This code appeared on page forty

At a different rate. A duplicate code at two prices is never quietly merged — one of the two is wrong, and only a person can say which.

That last one is worth pausing on. A duplicate code at two different prices is never quietly merged. It is raised, because one of the two is wrong and only a person can say which.

A human works through the flags — correcting, recapturing, deleting, adding anything missed. Typed corrections are final and the engine does not overrule them.

Approved pages are then frozen: locked, and not changed by anything that happens afterwards.

The part that matters every year, not just once

This is not a one-time problem.

Your suppliers issue new editions constantly, and today that means somebody compares two PDFs by eye, or nobody does and your rates quietly go stale.

Because the supplier's profile is saved, a new edition is re-read against it — and the output includes a comparison with the last one.

That single report is worth the exercise on its own. It tells you what to update, what to add to your range, and what to stop quoting — from a document you would otherwise have filed and forgotten.

Which prices changed

And from what, to what. The line you would otherwise have found out about from a customer.

Which items are new

The additions to your supplier’s range, so you can decide whether they belong in yours.

Which are discontinued

So you stop quoting what nobody can send you any more.

Who actually runs this

You send us the price lists.

We run them, check them, attach the images, and load the result into your product master.

Where a supplier's catalogue or website has usable images, those are used. Where the long tail has none, your own team can photograph the display.

Price list
Price list — Verma Tradersw.e.f. 01 Aug 2026
ProductMRPYour rate
Soft-close hingeFull overlay · 110° · clip-on More info₹ 214₹ 149
Drawer systemTandem · 450 mm · 30 kg load More info₹ 1,980₹ 1,386
Cabinet handleMatte black · 192 mm · zinc More info₹ 340₹ 238
Prices are subject to change. Every line carries the date it applies from.
Illustrative document. Figures are examples, not a price list.

One honest limit: this works on proper digital files — the PDF your supplier emailed you, or an Excel sheet.

A scan or a photograph of a printed page is not the same thing, and we will ask you for the original file rather than pretend otherwise.

Access to the tool itself, so you can run your own supplier price lists, is in development.

Send us one supplier price list and see what comes back.

Ask for a demo