What an RFQ Package Looks Like
A molding RFQ often arrives as an email with a STEP file, a PDF drawing with a title block, GD&T callouts and flag notes, a spec sheet naming the resin and grade, a volume forecast, and a supplier quality manual or quality agreement. Medical and automotive customers add requirements such as PPAP level, cleanroom molding, resin traceability or lot size limits. The estimator must find the dimensions that drive tool and process risk and then build the cost.
Much of the time is reading, not thinking.
- Resin grade and any approved alternates
- Critical-to-quality dimensions and tolerances tighter than normal molding capability
- Cosmetic surface, texture and gate location restrictions
- Annual volume, ramp, release pattern and packaging
- Secondary operations: ultrasonic welding, pad printing, insert molding, assembly
What the AI Extracts and Flags
A well-built agent turns the package into a structured summary with citations to the page and line where each fact came from. 05 mm on a long glass-filled nylon feature, an undercut that requires a slide, wall thickness variation likely to cause sink or warp, a requirement for cavity-level traceability, or a quality clause you have not accepted before. These flags become the checklist for the estimator and for a conversation with the customer.
Finding Similar Parts and Actuals
Past jobs are the best predictor of cost. The agent searches the ERP for parts with similar weight, resin family, projected area and complexity, then returns actual cycle times, cavities, scrap, press used and margin achieved. It can also surface customer-specific history such as quoted versus actual variance.
This is where most shops lose time, because the data exists in the job history of DELMIAworks, Plex, Epicor, SyteLine or a spreadsheet but is hard to query by similarity.
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.
Drafting the Quote and Keeping It Auditable
Using your quote template, the agent fills in resin, weight, proposed cycle based on comparable jobs, press rate from your table, scrap allowance, packaging and secondary operations, and generates a draft with a source for each number. The estimator adjusts cavitation, press and margin, and approves. Nothing is sent to the customer or written to the ERP without that approval, and the full chain of what the AI read and suggested is logged.
Netray builds this as a private agent on your infrastructure.
- Every number is traceable to a drawing note, ERP record or rate table
- Estimator approval before any quote is issued
- Logged inputs and outputs for audit
Data Security for Customer Drawings
Customer drawings are often covered by NDAs, and some molders also handle ITAR or CUI-marked part data for defense work. Sending those documents to a public AI service can breach both. Running open-weight models on your own servers keeps drawings inside your network, with access controls tied to who may see which customer.
Check your NDAs and flow-downs before choosing a tool, and have your own counsel confirm compliance.
When This Is Not Worth It
If you receive a handful of RFQs per month, mostly from the same customers for similar parts, an estimator will beat any setup effort. The value rises with RFQ volume, drawing variety and the number of people quoting. Test on 20 past RFQs first and compare the draft against what your estimators actually quoted.
Frequently asked questions
Can AI read injection molding drawings?
It can read text, title blocks, notes, tolerance tables and many dimension callouts, and it can analyze 3D models with geometry tools. Accuracy on complex GD&T and handwritten notes varies, so extracted values should always carry citations for the estimator to verify.
Will AI quote injection molding jobs automatically?
For simple repeat parts it can produce a draft with little editing, but most molders keep an estimator approving every quote because cavitation, press choice, risk and pricing strategy need judgment.
Is it safe to upload customer drawings to ChatGPT?
Often it is not, because NDAs, quality agreements and export rules may prohibit sharing drawings with third parties. Use a private deployment on your own infrastructure and check the contract terms with each customer.
If RFQ reading and history searching eat your estimators' week, we can pilot a private quoting assistant on twenty of your past RFQs and measure it against your actual quotes.