Comparison
Could ChatGPT just do this?
It is a fair question and we get asked it often. The short answer: ChatGPT is excellent at language and has no idea what is in your catalog. Quoting is a catalog problem wearing a language problem’s clothes.
Why the catalog is the whole difficulty.
The hard part is not reading the email
Both tools understand “panel 60x60 4000K and the usual driver”. Only one of them can tell you that maps to two specific SKUs in your catalog, at this customer’s price, with one of them out of stock.
A confident wrong answer is worse than no answer
Asked for a code it does not have, a general model will usually produce something that looks right. In quoting that is not a small error — it is a wrong price on a document a customer may accept.
Pasting the catalog in does not scale
It works for a demo with twenty products. With thousands of SKUs and dozens of pricelists you hit context limits, the data is stale the moment you paste it, and there is still no way back into the ERP.
Someone has to be accountable for the number
A quote is a commercial commitment. Per-line statuses and an approval step exist so a person can see what was decided and sign it off — a chat transcript is not that.
Side by side.
Written to be accurate about both. If a row looks unfair to ChatGPT, tell us and we will fix it.
| ChatGPT | Ettore | |
|---|---|---|
| Understanding a messy written request | Very good | Very good — same underlying capability |
| Knowing your product catalog | No access unless you paste it in | Indexed from your ERP, continuously |
| This customer’s agreed price | No access | Read from the pricelist in your ERP |
| Live stock | No access | Read at the moment the draft is built |
| When it cannot find an item | Tends to produce a plausible-looking code | Marks the line No match; unrecognized never defaults to Ready |
| Getting the quote into your ERP | You copy and paste it | Approved quotes are written back |
| Record of what was decided | A chat log | Per-line status and an audit trail |
| Approval before anything is sent | Whatever you remember to do | A human click on every delivery step |
| Drafting and rewriting prose | Better | Not what it is for |
Which one you actually want.
Choose ChatGPT if
- You want to rewrite an email, summarise a specification, or draft a reply
- You are exploring an idea rather than producing a priced document
- The work is occasional and does not touch your catalog
- You need something today, for free, with no setup
Choose Ettore if
- Requests arrive constantly and every one needs matching to a real catalog
- Pricing depends on which customer is asking
- A wrong SKU or a stale price has commercial consequences
- You want the approved result to land in your ERP without retyping
Questions about the comparison.
Do we have to stop using ChatGPT?
No, and we would not suggest it. Different jobs. Plenty of teams use a general assistant daily for writing and use Ettore for the catalog work — they barely overlap.
Couldn’t we just upload our catalog to ChatGPT?
For a few dozen products, honestly, yes. For thousands of SKUs across dozens of customer pricelists it stops working: the data is a snapshot that ages immediately, stock is not in it, there is no route back into the ERP, and you have no per-line record of what was decided.
Is Ettore built on a language model?
It uses one for the part it is good at — turning messy written requests into structured line items. The catalog matching, the pricing and the approval flow are ordinary engineering against your ERP, which is exactly why the answers can be checked.
What stops Ettore inventing a SKU the way a chatbot would?
Matches resolve against your actual catalog, and anything unrecognized defaults to Check match rather than Ready. That is architectural, not a setting someone can switch off.
Try the comparison yourself.
Take a real request, run it through ChatGPT, then send it to us. The difference shows up on the lines you would have had to look up.