What Is in an RFQ Package for Pumps and Valves
An inquiry from an EPC or end user typically includes a project specification, a pump or valve data sheet (for centrifugal pumps often following API 610 formats), a piping and instrument diagram excerpt, a vendor documentation requirements list, an inspection and test plan, and commercial terms. The estimator needs the duty: flow, head, fluid, temperature, specific gravity, vapor pressure, NPSH available, and for valves size, class, end connection, shell material, trim and leakage class. The same fact may appear on three documents with slightly different values, and exceptions to the customer's specification are buried in clarification lists.
What an Extraction Agent Produces
A model reading these documents fills a structured form: operating conditions, material requirements, applicable standards, accessories, painting and preservation, testing, and documentation. For each field it records the source page so a human can verify. It also separates what the document states from what it infers, and marks any contradiction between documents.
- Duty point, fluid properties and operating limits with source citations
- Materials by component and any required material class, plus special service flags such as sour or oxygen
- Referenced standards: API 610, API 598, ASME B16.34, ISO 9906 or ANSI/HI grades
- Required tests, witnessing, inspection hold points and documentation lists
- Deviations from your standard offering that need engineering review
Connecting the Extraction to Your Own History
Extraction alone saves reading time. The larger value is what happens next: the draft is matched against past quotes and orders for similar models and materials, pulling actual build hours, test time, casting costs and win or loss outcomes from the ERP and quote history. The estimator sees that the last three duplex 8-inch Class 600 gate valves took 40 percent more machining than standard, or that a certain requirement for witnessed hydrostatic testing was priced at zero.
The model is retrieving your facts, not inventing numbers, and every figure can link back to a source record.
Try it here, free
AI Quote Template and Prompt Set for Manufacturing
A fill-in manufacturing quote plus a reusable AI prompt that turns RFQ details into a structured cost build-up, with a mapping to common ERP quote fields.
Accuracy, Review and Failure Modes
Extraction is wrong sometimes: units are misread, a table is parsed out of order, a scanned data sheet is illegible, or a requirement is in an unusual place. Design the review step for that. Show the source snippet beside each extracted value, require estimator confirmation on key fields before the draft proceeds, and keep a log of corrections so you can see which document types fail most.
Never let extraction output flow directly to a customer or into the ERP unreviewed. A model that is 95 percent right still needs a person looking at the 5 percent.
Privacy, Export Control and When Not to Bother
RFQ packages often carry customer proprietary data, and some include export-controlled technical data, particularly for defense or nuclear-related work. Public cloud AI tools may be unacceptable under your contracts, so run the model on your own servers where document access follows your existing permissions. If you receive only a handful of RFQs a month, or your inquiries are mostly simple catalog items, the effort is not justified.
This approach pays where inquiry volume is high, packages are long, and estimator time is the limit on how much you can bid.
Frequently asked questions
Can AI read API 610 pump data sheets?
Yes, if the data sheets are readable PDFs, spreadsheets or clean scans. A model can extract the fields into structured form and cite the page. Accuracy depends on document quality and layout, so plan for estimator verification of key values, especially duty point, materials and the referenced standards, before the data drives a quote.
Will AI decide the pump selection or price?
It should not. Pump selection from performance curves and pricing should be done by deterministic tools and engineers, with AI supporting by extracting inputs and finding comparable past jobs. Keeping selection and pricing logic explicit means errors are traceable and the estimator remains accountable for the commitment.
How do I keep customer RFQ documents private when using AI?
Run document parsing, retrieval and the language model on servers you control, apply the same access rules your document system uses, and log every use. Avoid sending the documents to third-party APIs. Check your NDA and any export control obligations, since some terms prohibit disclosure to outside processors.
Send Netray a sample data sheet and RFQ package and see the structured extraction with source citations, produced on hardware you control.