Would You Let AI Negotiate With Your IT Supplier?

Yes, let AI negotiate with your IT supplier, but only on the parts of a deal that live in a spreadsheet. AI is excellent at comparing specifications, normalising quotations and drafting a clean request for quotation (RFQ); it is poor at the things that move an IT order in practice, such as supplier allocation, lead times, warranty nuance and the judgement call on which compromise is acceptable. Here is the supplier's-side view.

We use real listings from our store as examples, including the ASUS ExpertBook B1 B1403CVA-NJ1209X, the ASUS ExpertBook P3 P3406CCA-LY0040X, Microsoft 365 Business Standard and the ASUS Ascent GX10 for buyers who want to run the analysis on their own hardware.

What AI can and cannot do in IT procurement

Is anyone already letting AI negotiate with suppliers?

Yes, at the very large end of the market. In a January 2023 press release, the negotiation software company Pactum said that Walmart had deployed its autonomous negotiation technology with suppliers. These are Pactum's own, vendor-reported figures: the chatbot closed deals with 68% of the suppliers it approached, generated an average saving of 3%, and 83% of suppliers described it as easy to use. Pactum also said the tool let Walmart negotiate with 2,000 suppliers simultaneously.

The lesson is about scale. A retailer with thousands of contracts has a long tail no human team can revisit. A Singapore SME buying 20 laptops and a year of Microsoft 365 has the opposite problem: few suppliers, few line items, a lot of detail per line.

Where does AI help most when you buy IT?

Before the negotiation, not during it. The most expensive mistakes we see in IT purchasing start with an incomplete request, and AI is very good at filling the gaps in a request before it is sent. Ask a chat assistant to turn "we need 20 laptops for finance" into a structured RFQ and it will prompt you for specification, OS edition, warranty, delivery and payment terms. That is most of the checklist in our insider post What IT Vendors Wish Customers Knew Before Asking for a Quote, and a vendor answering that request can quote once instead of three times.

AI is also good at the tedious job of lining quotations up. One supplier quotes GST-inclusive, another quotes carry-in warranty; a model can put both into one table with a column per assumption. In Singapore, the IRAS rate page lists the current GST rate as 9%, so ask the tool to show every quote both before and after GST. Our guide Why the Cheapest IT Quotation Isn't Always the Best Deal shows the columns that matter.

Specification comparison works well too, provided the inputs are exact. The ASUS ExpertBook B1 B1403CVA-NJ1209X (Intel Core 7 150U, 16 GB, 512 GB SSD, S$1,499.00) and the ASUS ExpertBook P3 P3406CCA-LY0040X (Intel Core Ultra 7 255H, 16 GB, 512 GB SSD, S$2,099.00) look similar in a headline, and both listings state 3 years of onsite service from ASUS, next business day. A model given the two full spec sheets can explain the processor difference in plain language and ask which workloads justify the step up.

Where does AI get IT procurement wrong?

Three places, from where we sit.

Live supply. Prices and lead times in IT move with distributor allocation, currency, project pricing and end-of-life cycles. None of that is in a model's training data, and much of it is not public at all. An AI that confidently predicts "two-week delivery" is guessing.

Accuracy. OpenAI's own help centre says ChatGPT "can produce incorrect or misleading outputs", may sound confident even when wrong, and that users should "always verify quotes, data, technical information". A model that swaps a part number suffix or invents a warranty term will produce a comparison table that looks authoritative and is wrong. Part numbers are where IT orders succeed or fail, so check every one against the supplier's quotation.

Context. A good account manager knows your fleet's docks, ports and standard image. AI does not, unless you tell it, and it cannot weigh a substitute the way someone who has deployed both models can.

Use AI to write a better request, not to win an argument. The best price comes from a complete RFQ, not a clever chatbot.

Should I paste supplier quotations into ChatGPT?

Check the plan first. OpenAI states that content in consumer ChatGPT may be used to train its models unless the user opts out in data controls, while inputs and outputs from ChatGPT Business, Enterprise, Edu and the API are not used for training by default. Quotations often contain negotiated pricing and contact details, so use a business plan, follow your company's AI policy, or run the comparison locally. Our sibling post Your Company Data Is Feeding AI. Do You Even Know Where It Goes? covers the data question in more depth.

For teams that want analysis to stay on the premises, the ASUS Ascent GX10 (NVIDIA GB10, 128 GB, 1 TB, S$7,200.00 as listed) runs open models on a desk. It makes sense when procurement is one of several internal AI jobs; see our local AI vs cloud API cost model.

What happens when both sides use AI?

It is already plausible that an AI-drafted RFQ is answered with an AI-assisted quotation. That works when both are anchored to exact part numbers, and fails when each side assumes the other checked. Our view as a supplier: AI-written requests are welcome, and often clearer than hand-typed ones, but somebody on each side must own the numbers. The same theme runs through Why AI Projects Look Amazing in Demos but Fail in Real Offices, and the question of who signs off spending is the heart of Would You Trust an AI Agent With Your Company's Credit Card?

The case for and against an AI negotiator

For: it never forgets a line item, it is consistent and quick, and for organisations with many small contracts it can reach agreements no human team has time for.

Against: it cannot see supply, it can be confidently wrong, it may expose confidential pricing on the wrong plan, and it cannot build the relationship that gets you a call when stock is tight. For most SME purchases, a wrong part number costs more than automated haggling saves.

Our verdict: let AI prepare, compare and check; let people commit. Treat AI like a sharp junior buyer, as we discuss in One Employee, Ten AI Agents: excellent at the paperwork, supervised on decisions.

How to use AI to send a better RFQ

Ask the assistant to draft an RFQ with these fields: exact part number or minimum specification, quantity, operating system edition, warranty type and length, delivery address and required date, installation or imaging needs, software licences (for example Microsoft 365 Business Standard, listed at S$213.00 for 12 months), payment terms and, for public sector buyers, the procurement channel. Then ask it to list anything it had to assume, and answer those before sending. Government buyers can read our GeBIZ guide.

Buying IT from SourceIT in Singapore

SourceIT is an ASUS reseller and also supplies Microsoft 365 subscriptions. Prices above are as listed on 23 Oct 2026 and all four items were in stock at the time of writing. Code EXPERTBOOK10 gives 10% off the ExpertBook B1 and P3 14" until 30 Nov 2026. Browse notebook deals or software for business. Send your AI-drafted RFQ as it is: use Request for Quotation on the product page, call (65) 6978 3502, email Sales@sourceit.com.sg, see volume discounts and special pricing or use our contact page.

Frequently asked questions

Can AI negotiate prices with suppliers? Yes. Pactum says Walmart used its autonomous negotiation chatbot with suppliers, closing deals with 68% of those approached, by Pactum's own figures. For small IT orders, AI is more useful for preparing and comparing quotes than for haggling.

Can I use ChatGPT to compare IT quotations? Yes, for normalising GST, warranty and delivery terms into one table. Check every part number and figure yourself, as OpenAI advises verifying data and technical information.

Is it safe to paste supplier quotes into ChatGPT? Consumer ChatGPT content may be used for training unless you opt out. ChatGPT Business, Enterprise, Edu and API data is not used for training by default. Follow your company AI policy.

Can AI predict IT lead times? No. Lead times depend on live distributor allocation and stock, which public AI models cannot see. Ask the supplier to state lead time on the quotation.

What should an AI-written RFQ include? Exact part numbers or minimum specs, quantities, OS edition, warranty type and length, delivery address and date, installation needs, licences, payment terms and the procurement channel.