Sales Engineer
Know what the Sales Engineer is, where it lives in the platform and what it costs.
The Sales Engineer is Kabaido's flagship module, AI for industrial sales. It turns technical enquiries into configured quotes, purchase orders into orders and technical questions into cited answers, working from pasted text, attached files or a connected email inbox.
The problem it is aimed at
In industrial distribution and made-to-order manufacturing, the enquiry rarely arrives as a tidy list of part numbers. It arrives as a paragraph in an email, a scanned drawing schedule, a spreadsheet with three merged columns, a photograph of a handwritten cutting list. Somebody technical then reads it, works out what each line actually is, decides which of your products satisfies it, checks the price and writes the quote. That reading step is where the hours go, and it is the step that needs an applications engineer rather than an administrator.
The Sales Engineer does the reading. Not the deciding: the fit scores, the prices and the final quote all come from your own definitions and stay yours to approve.
Where it lives
New request in the main navigation is its front door, with search beside it. The Sales section holds what it produces: Quotes, Orders, Customers and Pricing. Through the email agent it also works natively in a Gmail or Microsoft 365 inbox; see the email agent page in Integrations.
What happens to a request
- Everything you sent, typed text and attached PDFs alike, is read into one document. Pages with a text layer are read as text; scanned pages are read by vision.
- A prescan picks which of the fifteen knowledge packs the content needs, so a bearing enquiry is read with power transmission vocabulary and a regrind list with cutting tool vocabulary.
- Typed line items come out: description, quantity, unit of measure, and the attributes (material, diameter, coating, tolerance) pulled into their own fields with units normalised and the original wording kept.
- Each line is resolved against what you sell, in a fixed priority: your catalogue first, then your saved designed products, then your services, configuring a fresh custom item where a configurator fits.
- The response is assembled: matches with their fit scores, clarifications for anything the request did not state, prices from your own rules, and the action card that creates the quote or order.
Two rules it will not break
Every extracted value carries a citation back to the exact span of your request it came from, so a figure on a quote can always be traced to the sentence or cell that produced it. And missing means a question, never a guess: a value the customer did not state stays empty and raises a clarification rather than being filled in with something plausible. A specification, a tolerance, a price or a part number is never invented.
The parts of the answer that have to be repeatable are not model output at all. Fit scores are a weighted calculation over your schema fields, the match bands are fixed thresholds, and prices come from one resolution function shared by the pipeline, the quote builder and the API. The same line against the same catalogue gives the same result every time.
It is useful before you import anything
Send a request with an empty catalogue and you still get the structured specification back: typed lines, normalised units, citations, clarifications. Matching is the part that switches on once your products are in. That order matters, because it means the module is worth something on day one rather than after a data project.
Where to read next
- Sending a request for what you can send and how it arrives.
- How Kabaido structures requests for the shape of the result.
- Working with matches for how fit is scored and how to overrule it.
- Application Engineering Intelligence for the knowledge model underneath.
What it costs
Included in every plan with no add-ons. Your plan's monthly usage meters the AI per resolved line; manual platform use is always free. See Plans and usage in Billing.