What a Forging RFQ Package Contains
A typical package has the finished part drawing, sometimes with a customer-supplied forging drawing showing parting line, draft angles and machining allowance. It lists a material specification such as an ASTM A182 grade, an AMS alloy specification or a MIL specification such as MIL-F-7190 for steel forgings, plus notes for heat treatment, mechanical testing, NDT, grain flow, marking and certification. A terms document may carry quality clauses, flow-downs and tooling terms.
The estimator has to reconcile all of it into one quote and catch anything that changes cost, such as a requirement for testing from a forging prolongation or a customer-approved source for material.
- Part and forging drawings with notes and tolerances
- Material and process specifications
- Quality, testing, NDT and certification requirements
- Terms, flow-downs and tooling ownership language
What AI Extracts Reliably
A language model with document parsing is well suited to pulling structured fields from a messy package. It can list the material grade and specification, quantities and delivery dates, heat treat condition, required tests and acceptance levels, marking requirements, and certification deliverables, with the page and note each came from. It can also detect spec references you may not stock, such as a revision you do not have on file.
Treat the result as a checklist for the estimator, not an authority: drawing notes are sometimes ambiguous, and the model should show its source so a human can verify in seconds.
- Grade, specification and revision with page reference
- Heat treat, test and NDT requirements as a checklist
- Quantities, release schedule and delivery terms
- Missing or conflicting information flagged for follow-up
Finding Similar Past Forgings
The more valuable step is matching the new part to history. Retrieval over your ERP and quote archive can search by material, weight range, forging process and customer, then show actual input weight, yield, press hours, scrap, die life and heat treat vendor for the closest matches. The estimator sees what a similar part really cost, not what it was quoted at.
This works only if the history is captured in the ERP in a form the system can read, and it works better when part families are consistently named. Geometry-level matching from drawings is much harder and less reliable, so expect the estimator to confirm similarity.
Try it here, free
AI Contract Negotiation Checklist
A structured checklist of pricing, data rights, liability, and exit terms to negotiate before signing an enterprise AI vendor contract.
Contract Terms and Export Control
AI can screen terms against a list of clauses your company has pre-approved and flag those that differ, such as unusual tooling ownership, liquidated damages, or flow-downs that require specific certifications. It does not give legal advice, and approval stays with a person. If the package contains export-controlled drawings under ITAR or EAR, run the AI on infrastructure you control rather than uploading to an outside service, with user access limited to authorized persons.
Netray's approach is private models on your own hardware, which lets you apply your existing access controls rather than create a new data path.
- Compare terms to your approved clause list
- Run on-premises when drawings are export-controlled
- Keep estimator and contract reviewer approval mandatory
When Not to Bother
If you receive fewer than a few RFQs a week, or if your customers send clean, repeat requirements, reading them yourself is quick and the setup effort will not pay back. AI review also adds little if your past job data is incomplete, because the most valuable output is the comparison with real history. Pilot on a month of old RFQs and measure whether the extracted checklist matches what your estimators would have written.
Frequently asked questions
Can AI read forging drawings?
It can reliably read text on drawings, such as notes, material callouts and specification references, and increasingly some dimensions. Judging forgeability, die complexity or grain flow from geometry still needs an experienced estimator.
Is it safe to upload ITAR forging drawings to an AI tool?
Not to an uncontrolled external service. Export-controlled technical data should stay inside infrastructure that meets your compliance program. Running models on your own hardware avoids creating a new third-party data path.
Will AI replace forging estimators?
No. It handles reading and retrieval so estimators spend time on judgment: yield assumptions, die design, risk and pricing. Anything affecting price or commitments should be approved by a person.
Send Netray a sample forging RFQ and see the extracted checklist and similar-job comparison built on your own infrastructure.