Ask most manufacturing engineers how long an RFQ takes from drawing to quotation, and the honest answer is usually “it depends” — followed by a description of a process with several manual handoffs, each one adding time and risk. This is a breakdown of where that time actually goes, and what it looks like to connect the steps instead of leaving them disconnected.
Mapping the manual version
A typical RFQ, handled manually, looks something like this:
- A drawing arrives — by email, usually as a PDF.
- An engineer opens it and reads the title block by eye: part name, drawing number, revision, material, weight. This gets typed into a tracking spreadsheet.
- Someone (often the same engineer, sometimes a dedicated reviewer) balloons the drawing — numbering every dimension and tolerance — by hand, on a printed copy or in a separate markup tool.
- Costing takes the part data and rebuilds it a second time in a costing spreadsheet, calculating material, tooling and process costs, often by referencing the balloon sheet manually.
- The quotation is built from the costing output, re-entering customer and part details a third time into a quotation template.
- If approved, the quotation goes out — and somewhere, a customer record gets updated manually, if it gets updated at all.
Every arrow in that sequence is a manual re-entry step. Each one takes time, and each one is an opportunity for a transcription error — a dimension typed wrong, a material grade that doesn’t match between the drawing and the costing sheet.
What “automating” the RFQ actually means
Automating this doesn’t mean removing the engineer from the loop — it means removing the re-typing between steps, so data entered or verified once carries forward automatically, with a human checkpoint at the stages that need judgment.
Concretely, that looks like:
- Drawing intake: title block fields (part name, drawing number, revision, material, weight) extracted automatically from an uploaded drawing, with a readiness check flagging anything missing.
- Ballooning: every dimension and tolerance numbered and plotted automatically across every sheet, with a confidence score on each one — Verified, High, Review, or Manual — so an engineer’s attention goes to the fields that actually need checking.
- Review: a dedicated workspace for zooming into the drawing, correcting a misread value, or re-cropping a view — the human checkpoint that keeps this from being a black box.
- Process route: confirmed by an engineer, not inferred, so costing has an accurate route to calculate against.
- Costing: material, tooling amortisation, machining and surface-treatment costs rolled up automatically from the balloon data and process route — editable line by line before anything is finalized.
- Quotation: generated from the same data, with customer and part details carried through rather than re-typed.
- Pipeline and CRM: once approved, the RFQ moves through a tracked pipeline with role-based sign-off, and a customer record is created or updated automatically.
A working example, not a concept
This isn’t a hypothetical workflow — it’s how InFaktory is actually built, as a seven-step AutoQuote wizard for die-casting RFQs specifically, reading real casting and machining drawings and carrying the data through to a costed, exportable quotation. The confidence-scoring and mandatory review step exist because drawing extraction isn’t perfect, and the system is built around that reality rather than pretending otherwise.
Where the real time savings come from
The time saved isn’t primarily from the AI extraction step being fast, although it is — it’s from eliminating the two or three redundant re-entry steps downstream of it. An engineer who reviews and corrects an auto-generated balloon plot is usually faster than one building the same plot from scratch, and costing that pulls from already-verified balloon data is both faster and more consistent than a parallel manual rebuild.
Why this matters beyond speed
Faster RFQ turnaround is the visible benefit, but consistency is the less visible one. When costing always pulls from the same verified source data instead of being rebuilt by hand each time, pricing becomes more consistent across quotes — which matters as much for margin protection as it does for customer-facing speed. See Manufacturing Quotation Software for more on the costing side of this specifically.