Short answer: by combining three sources. The AI reads what the customer wrote, compares it with your item master using fuzzy matching on names, codes, sizes and synonyms, and weighs the customer's own order history to see what they usually buy. The best match is proposed with alternatives, and a person only steps in when the AI is unsure. This is what removes the research work behind manual order entry.
Why this is the real bottleneck
Typing an order takes a few minutes. Finding out which item a customer means can take much longer: searching the ERP, checking old orders, sometimes calling the customer. In manufacturing and wholesale, with thousands of items and customer specific names, this is where most time goes.
How the matching works
1. Fuzzy matching on product data
Small differences in spelling, word order, units or abbreviations do not stop the match. "RVS bout M8x40" and "stainless bolt 8 x 40 mm" end up at the same item.
2. Customer order history
If a customer always orders a specific variant, that variant gets priority. References such as "same as last time" become usable.
3. Customer codes and instructions
Many customers use their own article numbers. ENTR learns these cross references and lets you add instructions in plain language.
| Aspect | Manual lookup | With ENTR |
|---|---|---|
| Vague description | Search ERP, ask a colleague | Best match proposed instantly |
| Customer article codes | Separate lists or memory | Cross references applied automatically |
| "Same as last time" | Dig up the old order | Uses order history |
| Uncertain match | Guess or call the customer | Flagged with alternatives |
| Knowledge in the team | Lost when people leave | Kept in the system |
Result in practice
Many manufacturers receive orders and quotation requests as raw email text. With order history and fuzzy matching the team no longer has to research which items are meant, which is where most of the time saving comes from.
Conclusion
Reading text is solved. The value is in translating what the customer means into what your ERP knows. Ask any vendor to show this with your own item data.
