How I Use AI to Compare Jewellery Suppliers (Without the Spreadsheet That Never Gets Finished)

You know the spreadsheet. You open it with good intentions. You paste in a few quotes, start a column for lead times, add another for minimums. Then a client emails. Then you forget where you put the spec sheet from that caster in Hatton Garden. Three weeks later the spreadsheet has five rows and a note that says “check this later.” The quotes are already out of date.

I have been there. I run Silux London alongside the Maker's AI Lab, and supplier comparison used to eat a half-day every time I needed to refresh a vendor relationship. Then I started using AI the way you would use a methodical assistant, not just a fancy search bar. The difference was significant.

This post shows you the method, step by step. No coding. No technical background needed. Just a clear prompt and the AI chat tool you probably already have open.

The core idea: a small team, a clear scorecard

Most people use AI in one big conversation. They paste things in, ask questions, read the answer, paste more things in. That works, but it is slow and easy to lose track.

The smarter way is to think of the AI as running a small tournament on your behalf.

Here is how I describe it to students at the academy. Imagine hiring a research assistant who is very fast but not very experienced. You give them a pile of quotes and spec sheets and you say: “Score each of these against these five criteria. Give me your best two. Then I want you to double-check the top two before you hand it to me.”

That is exactly what you can do in a single AI session. The AI processes each supplier one at a time, against criteria you decide in advance. It surfaces the top two or three. Then you ask it to go back and check its own reasoning on those finalists. Then you make the call.

The human still decides. The AI handles the tedious scoring.

A worked example: comparing ten casters

Let's say you have collected quotes and information from ten casting houses. You have emails, a few PDF spec sheets, and some notes from trade shows. Some are local, some are overseas. You need a caster for a new silver collection, hallmarking is non-negotiable, and you need a turnaround of under three weeks.

Here is how to run the comparison in one AI session.

1. 1
   Gather everything into one place.

   Copy the relevant information from each supplier into a single document or note. You do not need it to be tidy. Name, price per gram or per piece, minimum order, stated turnaround, hallmarking (yes or no, which assay office), and any notes on communication quality. Paste the whole thing into the AI chat.

2. 2
   Define your criteria before you ask for a ranking.

   This is the most important step. Tell the AI what matters to you. Do not let it guess. Different makers weight things differently: a hobbyist might prioritise small minimums; a brand selling wholesale needs reliability of hallmarking above everything else.

3. 3
   Use this prompt (copy and adapt it).

   Copy-paste promptBelow is information I have gathered on [number] casting suppliers. Please score each one out of 10 on these criteria: 1. Price competitiveness (based on the figures I have given you) 2. Turnaround time (my target is under three weeks) 3. UK hallmarking compliance (non-negotiable: score 0 if not confirmed) 4. Minimum order quantity (I need to be able to order [X] pieces per run) 5. Communication quality (based on my notes) Give each supplier a score for each criterion. Weight hallmarking compliance the highest: it counts double. Then give me an overall score and a one-sentence summary for each. List them from highest to lowest. Here is the supplier information: [paste your supplier notes here]Copy
4. 4
   Ask it to double-check the top two.

   Once it gives you a ranked list, pick the top two or three and ask:

   “Look again at [Supplier A] and [Supplier B]. What did I not give you enough information on? What would you want to verify before I commit to either of them?”

   This mimics the verification step a good research assistant would do. The AI will flag gaps: “You did not mention whether Supplier A uses an approved London assay office or Birmingham,” or “The minimum order for Supplier B seems high for the volume you described.” These are exactly the questions you need to ask before you sign anything.

What you still decide

The AI is scoring against numbers and notes you provided. It cannot assess trust. It does not know that the caster you used three years ago for a rush order went above and beyond. It cannot hear the tone of an email or know that a supplier quoted low last time and then added charges later.

A few things that are always yours to judge:

• Relationship history. If you have worked with someone before and it went well, weight that.
• Gut feel on communication. A supplier who takes four days to reply to a quote enquiry may take longer on a production question.
• Verification of compliance claims. Hallmarking and ethical sourcing claims need to be checked. Ask for documentation. The AI cannot do this.
• The actual visit or sample. For a new casting partner, request a sample piece before committing to a full run.

Use the AI's ranked list as a shortlist, not a final answer. It does the legwork; you make the call.

Where else this pattern works

The same method, with different criteria, applies across a lot of the decisions that pile up when you run a small maker business.

• Ranking stone vendors. Paste in quotes for the same stone spec from three or four vendors. Score on price, certification (Kimberley Process, GIA grading), minimum spend, and delivery time.
• Triaging wholesale enquiries. When you get multiple stockist enquiries at once, paste them all in and score against your ideal stockist profile (order size, location, brand alignment, payment terms).
• Choosing which trade shows to attend. List the shows you are considering with their costs, locations, audience profiles, and dates. Ask the AI to score each one against your goals for the year. It will not know the shows deeply, but it can force you to articulate your criteria, which is half the work.
• Comparing packaging suppliers. Same method: paste the quotes, define what matters (materials, minimum quantity, lead time, UK-based or not), get a scored list.

The pattern is always the same. Define criteria first. Feed in data. Get a ranked list. Ask it to flag what is missing. Then decide.

A note on accuracy

Want to go further?

If this post gave you one useful idea, the next step is to practise it. Open your AI chat tool today and try it on something small: three packaging suppliers, two stone vendors, anything where you have been putting off a comparison because it feels like too much work.

If you want this kind of thinking woven into how you run your whole business, come and join the community. It is a membership for makers and brand owners learning to use AI in their craft, with prompts, practical walkthroughs, and other makers working on the same problems. It is always open, so you can start whenever suits you and learn at your own pace.

I also run deeper, guided programmes from time to time for makers who want to go further. Members hear about those first.

You will find it all at hamedarab.academy.