To compare supplier quotes that arrive as PDFs, Excel files, scans and email bodies, fix one comparison template first: the same items, units, delivery terms and payment terms for every supplier. AI can then read each quote into that template and flag gaps and unit mismatches, while a buyer checks every line before anyone picks a supplier.
#Key takeaways
- Most of the work in a quote comparison is normalising, not deciding: units, pack sizes, delivery terms, validity and exclusions.
- A fixed comparison template, sent out with the RFQ, removes more work than any tool.
- AI is good at reading mixed formats and languages into that template and pointing at what is missing.
- The award, and every number that feeds it, stays with a named buyer.
- If suppliers already answer inside your sourcing module, a spreadsheet or that module is enough.
#What does quote comparison look like in a normal week?
A buyer sends an RFQ for packaging materials to five suppliers. Replies come back as a PDF on letterhead, an Excel file with its own column order, a scanned page with handwritten notes, an email body in Arabic and a WhatsApp photo of a printed quote. One supplier quotes per carton, another per pallet, a third per unit, and one quietly excludes delivery.
The buyer opens each file, retypes lines into a sheet, converts units, chases the supplier who left out the validity date, and only then compares. Most of the effort goes into translating quotes into a common shape, not into judging them.
It is a common load. In the Procurement Tactics and Suplari AI readiness survey, checked november 2026, 72.1% of procurement teams described their manual workload as high or very high, much of it gathering and reshaping data.
#What should every quote be normalised to?
Before any tool, agree the template. A usable comparison has one row per item and the same columns for every supplier:
- The item, matched to your own item code, not the supplier's description.
- Unit of measure and pack size, converted to the unit you order in.
- Quantity breaks and minimum order quantities.
- Delivery terms (incoterms), delivery point and lead time as the supplier states it.
- Payment terms and how long the quote stays valid.
- Exclusions, assumptions and deviations from your specification.
- Missing fields, listed per supplier so the buyer can chase them.
Put this template into the RFQ itself. A supplier who fills in your format saves the most work of all, and a model that reads into a fixed template is far easier to check than one asked to 'compare these quotes'.
#What should AI do, and what stays with the buyer?
Let AI do the reading: extract each line from PDFs, spreadsheets, scans and email bodies, in Arabic or English, into the template; convert units with a conversion table you approve; and flag what does not fit, such as an item it could not match, a missing validity date or an exclusion buried in the notes.
Keep the judgement with the buyer: which deviations are acceptable, whether a lead time is credible, how quality history and the relationship weigh in, and the award itself. Every extracted line should point to the page it came from, so the buyer checks the source in one click instead of trusting a summary.
Narrow, checked steps are the ones that last. Gartner, july 2025 placed generative AI for procurement in the trough of disillusionment, with workflow automation among the uses now emerging.
#When is a spreadsheet or your sourcing module enough?
If you send a handful of RFQs, suppliers reply in your template and quotes are mostly in one language, a well-built Excel template with locked columns is enough. If you already run sourcing events in SAP Ariba, Coupa or your ERP's RFQ module, push suppliers to answer inside it. Structured responses beat any extraction.
AI earns its place when volume is steady, formats are mixed, suppliers will not use your portal and quotes arrive in several languages.
#How does Stennir run this?
Stennir starts with AI consulting when quote comparison is one of several candidate workflows: the AI Opportunity Audit ranks five workflows, each with a cost and a payback case, and recommends one build. If quote comparison wins, AI implementation builds it in your own accounts: extraction into your template, unit tables, gap flags and a buyer approving every line.
Arabic and English are fully supported, other languages on request. After go-live, Managed AI Operations tracks extraction accuracy on real quotes and fixes what drifts, with a named owner. AI training then works with your buyers on the real workflow, so the checked comparison becomes the habit.
We are
Stennir builds the reading and normalising step around your own template and approval rules, with a buyer signing off every comparison.
We aren't
Stennir is not a sourcing suite, and not a tool that picks suppliers or sends awards on its own.
#What should a procurement manager do next?
Take your last three RFQs and write down where the time went. If most of it was retyping and unit conversion, fix the template first, then test extraction on those real quotes. See AI for operations and the purchase-order backlog worked example, then take the free readiness assessment or book a 30-minute call.
#Common questions
What do you need from our procurement team?
A set of recent RFQs with the supplier replies, your comparison template or the one you wish you had, your item codes and unit conversions, and a named buyer who approves output. The people who compare quotes today walk through real cases so the build matches how the work is really done.
Can it read quotes in Arabic or in other languages?
Arabic and English are fully supported, including scans and files that mix both. Other languages, such as Chinese or Turkish, are supported on request and tested on your own supplier files first. The buyer sees the original next to each extracted line, so a wrong reading is caught before it reaches the comparison.
Will it choose the supplier for us?
No. It reads, normalises and flags. The buyer decides which deviations are acceptable and who wins, and the award goes out under a person's name. Every extraction, edit and approval is logged, so an auditor can see who checked which quote and what changed along the way.
Where do our supplier quotes stay?
Where your rules require. Hosting is agreed in writing before the build: in your cloud, your country or your existing systems. The workflow runs in your accounts, so the quotes, the logs and the off switch stay with you, and you decide what any model provider may retain.
