Answers

Frequently asked questions

Everything a purchase head asks before letting an agent near a live enquiry — what it computes, what it will not do without a human, what your suppliers have to change (nothing), and what a seat costs. Where the honest answer is that a number depends on your plant, it says so.

01

What it is, and whether it fits us

The three questions a purchase head asks in the first five minutes.

  • Q01

    What does ProcureOS do that our Excel comparative statement and email chain does not?

    Two things a spreadsheet cannot. Every RFQ gets an evaluated comparison that lands each vendor on the same footing — unit basis, freight, payment terms, GST position — and then says who is actually lowest, per line as well as on the total. And before the first quote arrives you get a should-cost build-up drawn from your own purchase history, so you walk into the negotiation with a number you can defend line by line. Around both, the case, the clocks, the approvals and the PO live in one record instead of seven tabs and a WhatsApp group.

  • Q02

    Do we have to change how the purchase department works?

    No. ProcureOS takes the sequence you already run — indent, RFQ, quotes, comparative statement, approval, PO, delivery — and gives each step a place to live, a clock and a record. Your own rules go in as playbook rows: value bands, approval steps, reminders, skip rules, vendor routing. Nothing is automated or granted by default, and every version of the playbook is kept, so you can read what the rule was on the day a case was approved.

  • Q03

    We already run SAP / Tally / an ERP. Does ProcureOS replace it or sit alongside it?

    It sits alongside. ProcureOS is not a book of accounts and does not try to become one — it runs the stretch most ERPs are weakest at, from enquiry through quotes and evaluated comparison to award. What comes out the other end is a released PO with the awarded lines, rates and terms on it, which you post into your ERP the way you do today. There is no ERP connector at present; vendor and item master data comes in by CSV import.

02

Getting started

What it actually takes to run your first enquiry through it.

  • Q04

    How long does it take to go live, and do we have to clean up our vendor and item master first?

    You do not clean up first — that project never finishes and it is not a prerequisite here. Bring the vendor list in as a CSV (name, emails, GSTIN, PAN, MSME status) and add items as the cases come; nothing has to be configured before the first RFQ goes out. The master then improves as you buy, because every award writes a price point back against the item, which is exactly what the costing engine reads later.

  • Q05

    Can we run one live enquiry through ProcureOS before we commit?

    Yes, and it is the sensible way to buy this. Pick one live enquiry with three or more vendors quoting, run it end to end in ProcureOS while your team runs it the usual way in parallel, and put the two comparative statements side by side. The number worth arguing about is the gap between the lowest quoted price and the lowest evaluated price — if there is no gap on your enquiry, you have learned that cheaply.

03

The two engines

The comparison and the cost sheet are the reason the seat is worth paying for. Both are computed, not generated.

  • Q06

    How is the evaluated comparison different from the comparative statement my team types today?

    Yours is typed from what each vendor wrote on their letterhead. Ours is computed after every quote has been adjusted onto the same footing, so the vendor with the low headline rate and 100% advance does not beat the one offering 45 days credit by accident. L1 is flagged per line as well as on the total, which is how a split award becomes visible when it is genuinely cheaper. It runs itself the moment the quotes are in — nobody has to ask for it, and the same quotes always produce the same deck.

  • Q07

    What exactly gets normalised across vendors — freight, GST treatment, payment terms, quote validity?

    Four adjustments in a fixed order, because each compounds on the one before it. Unit basis first, so a rate quoted per set of two is divided down to your unit; then freight, allocated across lines by value where the vendor did not price it in; then payment terms, comparing the credit offered against what the RFQ asked for and valuing the difference at your cost of capital; then GST, because a composition or unregistered vendor cannot pass on input credit, so their tax is added to the evaluated price. Quote validity is tracked and shown, never guessed. A commercial term the vendor left unstated is recorded as a gap and chased — and a vendor carrying an open gap is blocked from the ranking entirely rather than quietly assumed into L1.

  • Q08

    Where does the should-cost figure come from, and what happens on an item with no purchase history?

    From your own records, in order of preference: the last confirmed cost sheet for that item, re-anchored to current index rates; failing that, your price-point history for the item; failing that, stated assumptions. Every line names which of those it came from, and a line only reads as market-backed if it has a live index anchor or three recent price points from three different vendors — everything else is presented as your history. On an item with no precedent at all the engine does not invent a number: it marks the lines it assumed, reports them as open assumptions, and refuses to let the sheet be confirmed while any of them is still unanswered.

  • Q09

    If a figure on the deck is wrong, can we correct it — and does the correction hold for the next RFQ?

    Yes, and yes. Say it in the thread — the steel rate is stale, that freight is inclusive, this line is per set — and the deck or cost sheet recomputes as a new version with the old one kept alongside it. The correction is also stored as an example against your organisation, and once promoted it seeds the same line the next time you buy that item, so your team starts from the number a human already fixed instead of making the same edit every quarter.

04

Our suppliers

The part that usually kills procurement software: nothing here asks a vendor to change anything.

  • Q10

    Do our suppliers have to register, log in, or learn a portal?

    No, and there is no plan for them to. Suppliers have no seat, no password and no portal — they receive the RFQ as ordinary email and reply the way they always have. There is nothing for your team to onboard them onto, which also means there is nothing for them to refuse. Suppliers are never charged either.

  • Q11

    What happens when a vendor replies with a photo of a quote, a scanned PDF, or “same rate as last time”?

    The reply is read on arrival and then echoed back before it counts. Nish stages a short confirmation email that reads the commercial terms back to the vendor in words — rates, quantities, freight, GST treatment, credit days, validity — and the quote does not enter the comparison until they confirm it. Anything unreadable is quarantined rather than guessed at, and a quote that arrives by phone or over the counter is entered by hand as an offline quote. A newer quote from the same vendor on the same RFQ automatically supersedes their earlier one, so you are never comparing against a stale price.

  • Q12

    Does ProcureOS email our suppliers on its own, and whose address does the RFQ come from?

    It never sends on its own. Every external email — the RFQ, the reminder, the echo-back, a target price, the PO — is drafted and queued as a staged email that sits there until a person on your team clicks Send. The mail then goes out from ProcureOS on your purchase department's behalf, carrying your case reference in the subject line so the vendor's reply files against the right enquiry. Nothing has ever reached a supplier without a human deciding to send it.

05

Seats, roles and approvals

One paid role. The rest of the plant joins for free.

  • Q13

    Who needs a paid seat and who uses ProcureOS free?

    Only purchasers hold a paid seat — that is the one that carries the shell, both engines, awards, PO approvals, the playbook, roles and reports. Finance is free and works from the inbox. Requesters on your work domain are free and join automatically, and they can raise and track their own requests. Suppliers never hold a seat at all.

  • Q14

    What does a purchaser seat cost?

    We do not publish a figure, because a two-purchaser job shop and a six-plant group are not the same purchase department and a list price would be wrong for both. What is fixed is the shape: you pay for purchaser seats and nothing else. Finance seats, requester seats and supplier access are free, and there is no per-RFQ, per-PO or per-vendor charge sitting behind the number. Tell us how many purchasers and how many plants and we will quote for it on a call.

  • Q15

    How do approvals work if our finance head never logs into software?

    They never have to log in. The approval arrives as an email with the numbers already inside it — the case, the vendor, the amount, the budget head and what is left on it — and they approve or reject from there. Shell access is available if they ever want it, but nothing in the process waits on them signing in. If an approval sits too long, the clock chases it on the schedule your playbook sets.

06

Trust, data and control

What Nish is not allowed to do, and what happens to your records.

  • Q16

    Can Nish award an order, release a PO, or send an email to a vendor without a human?

    No. Nish drafts, computes and chases; a person approves the budget, awards the order, releases the PO and clicks Send on every external email. There is no auto-award switch to leave on by mistake — it is not a setting that happens to be off, it is a capability the agent does not have. Drafting is automated and drafts can be wrong, so the supplier's own quote and documents govern over any summary of them.

  • Q17

    Can a requester — or anyone outside the purchase department — ever see prices?

    No, from any surface. Ask “where is my bearing order?” as a requester and you get status and dates; the prices are simply not in the answer. Retrieval is scoped to the role before the records are read, so this is not a matter of hiding a column in the interface. Every row also carries your organisation and is read back through that scope on every query, so one plant's prices are not reachable from another's account.

  • Q18

    Do you train AI models on our quotes and vendor list, and what happens to our data if we stop using ProcureOS?

    We do not use your uploaded documents to train general-purpose AI models, and Enterprise customer content is not used for product improvement or model training at all. The corrections your team makes are stored against your organisation and improve your decks and cost sheets, not somebody else's. Retention, export and what happens on cancellation are set out in the Privacy Policy and the Terms — read those two rather than this box, because those are the documents that bind us.

Still deciding

Ask the question that is not here.

These are the eighteen we are asked most. If the one blocking your decision is a nineteenth, write to the desk that owns it — a person answers, and the answer will be as specific as the ones above.

Buying it

sales@trynishai.com

Pricing for your plant, or setting up a live enquiry to run in parallel.

Already using it

support@trynishai.com

A deck that read a quote wrong, a vendor who is not receiving RFQs.

Your data

privacy@trynishai.com

Access, export and erasure requests under the DPDP Act.