Skip to content
Daniel Östling.

RFQ-to-Quote — custom build for manufacturing clients

Multimodal prototype that cuts a 30–60-minute estimator workflow to under one minute.

Founder / built end-to-end · 2026

PROBLEM
CPQ software can't read drawings, PDFs, or email RFQs, leaving estimators to extract specifications manually.
BUILT
A multimodal prototype that reads technical drawings and RFQ emails, extracts specifications, and drafts customer clarifications.
OUTCOME
30–60 min estimator workflow · < 1 min prototype workflow · 3 prospective-company demos
STATUS
building
The system reading a real bearing-housing drawing — pulling the full spec, judging what's quotable, and pinning its questions back on the drawing.

Custom-product businesses — print, packaging, signage, configured goods — quote before they sell. The quote starts as a messy PDF, a technical drawing, an email thread. CPQ software can’t read any of that: it handles structured catalogs, not drawings. I’m building the layer that can: a vision-model system that reads the RFQ as it actually arrives and drafts a line-item quote for an estimator to review and send.

What it reads

The inputs are rarely clean. A drawing can be sharp CAD output or a 40-year-old hand-marked scan, in German or Norwegian, with the detail that decides the price squeezed into a margin. The system reads both the same way.

A decades-old, hand-marked engineering drawing of a bearing housing
A decades-old, hand-marked bearing-housing drawing — the kind the system reads in seconds.
An engineering drawing of a multi-position needle rod with tolerances and surface-finish annotations
A multi-position needle rod — the case where the AI flagged the same clarifications the estimator did.

How it works

Drop in a drawing — PDF or image — and the system returns the full specification in seconds, cross-references the customer’s email, and flags exactly what a senior estimator would question.

  1. Drawing extraction. It reads geometry, key dimensions, materials, tolerances, surface finishes, thread specs and bolt-hole patterns — from clean CAD to decades-old hand-marked scans — and returns a structured specification as part of the under-one-minute workflow.
  2. Context from the RFQ email. The customer’s email is parsed alongside the drawing — quantities, delivery dates, material preferences — and cross-referenced against what the drawing actually shows, so nothing falls through the gap between the two.
  3. Ambiguity detection. It flags what a senior estimator would catch — missing quantities, contradictory specs, superseded standards, the judgment calls that need a human — each categorised by why it matters.
  4. Clarification email, drafted. Where the customer needs to answer a question, the system drafts the email — in the customer’s own language and the shop’s voice, ready to send with a light edit.
  5. Quote assembly. Part details flow straight into the shop’s own quote template — its layout, its branding — pre-filled from the extraction and ready for pricing and export.

What the prototype proved

The prototype reduced a 30–60-minute estimator workflow to under one minute. An experienced estimator validated the output, and I demonstrated the working prototype to three prospective companies.

Status

In build, working with real RFQ data after validation and three prospective company demonstrations. If you run a quote-first business and want to see it on your own RFQs, email me.

Want something like this? Email me →