Pricing
Work out your cost
You buy credits and spend them only on tenders you actually run — no seat licences, no minimum commitment, and nothing charged for a quiet month. Set the numbers below to match your procurement and see the estimate.
Everything below is per tender; this multiplies it.
The tender's own RFP. Layout is read once per page, so this is linear in RFP length.
Typically 50–150.
Every page of every submission is read once.
The biggest lever: each bidder adds both pages to read and checks to run.
Bidders decide this, not you.
Each criterion is checked against each bidder separately. The 3-vote loop is already inside the rate.
PQ, technical and commercial together.
Drafting wizard, pre-bid and corrigendum assistants, clause Q&A — and officer re-checks, which run the full agent again.
Wizard, pre-bid replies, corrigendum drafting, Q&A.
A re-check costs a full criterion check, not a chat turn. 100% would double the evaluation stage.
How the estimate is built
What is a credit?
A unit of AI work — reading a page of a submission, or checking one criterion against one bidder, each cost a set number of credits. The rates come from measuring the platform doing real tender work, not from an estimate. The panel above lists them.
Why is this not a fixed monthly fee?
Because a quiet month would cost you the same as a busy one. Credits are spent only when a tender is actually set up, read or evaluated — no seat licences, and nothing charged for a month you do not run a tender.
Do bidders need credits?
No. Firms submitting to your tenders are never charged and need no account credits, so nothing about your pricing discourages a competitive field. Reading their documents is covered by your balance.
What if more bidders apply than I expected?
The estimate rises, because every submission is read and every criterion is checked against every bidder. The panel above shows exactly what one extra bidder costs on one tender, so a larger field is never a surprise — and a smaller one costs proportionally less.
How accurate is this estimate?
Reading pages is very steady — measured across thousands of real pages, the cost of a page varies by under ten percent. Evaluation varies more, because a criterion whose evidence is scattered across a submission takes more work than one answered on a single page. Treat the reading figure as reliable and the evaluation figure as an average.
How do you count a page?
A page is a page of a submitted document, as it was submitted. The platform reads the page as an image, so a page of stamps, tables or scanned certificates costs the same as a page of text. A typical tender page carries around 181 words, which is lower than a prose page because most of a submission is forms and tables.
Can I see what a tender actually used?
Yes. Consumption is metered per job and itemised in your account, so an estimate is only ever a starting point — once a tender has run you can see what it really cost.
Per-unit rates are derived from measured platform usage and may be revised; existing credit balances are never repriced.