Manufacturing /Document Generation live field guide · 8 min

RFQ-to-Quote Copilot

Parses an incoming RFQ — email, web form or PDF — into materials, quantities and specs, checks them against your bill-of-materials and lead times, and drafts a structured quote, flagging spec gaps for the estimator to resolve before it reaches the customer.

Open the live demo

theater/demos/manufacturing_rfq-copilot.html · sandbox · read-only

Open

FIG. 1

The live demo, running on fabricated data. Open it to step through the full flow — every output is shown for a person to approve before anything happens.

How it would work

Reads the incoming RFQ and its attachments, matches each requirement to your BOM and lead times, drafts a priced quote with every line traced to its source, and lays it out for an estimator to correct and approve before it reaches the customer.

Input 01

The RFQ and its attachments

An incoming RFQ — email body plus drawings, spec sheets and spreadsheets — with part, quantity, material, finish, tolerances, delivery date and any special terms scattered across the pack.

Agent 02

Extracts, matches, prices

Pulls each requirement out of the pack and tags its source, matches the part to your BOM and current lead times, prices material, labour, finishing and freight, and flags anything it could not match cleanly.

Output 03

A draft quote, working shown

A priced draft with each line traced to where it came from, an extraction confidence score and any risk flags, laid out for an estimator to adjust margins, fill the judgement calls, and approve before it goes to the customer.

Where it works well

It does the tedious extraction every time, on every attachment, and shows where each figure came from.

Done by hand a standard RFQ is half a day of reading and re-keying before pricing even starts — and the spec buried on page three is the one that gets missed.
Best for estimators and sales engineers fielding a steady flow of RFQs: sheet metal, machining, fabrication, plastics.
At dozens of RFQs a week, the recaptured hours go back into the judgement calls and the non-standard jobs, not the data entry.

The slow, invisible cost of quoting is reading the pack — pulling material, quantity, finish, tolerances and a delivery date out of an email body, a drawing revision and a spec spreadsheet, then keying them into a quote without losing one.

Where it works badly

It is confidently wrong when a spec is ambiguous or your BOM and lead times are stale.

Weak where a requirement doesn't map cleanly to one BOM line — a custom or quotable item that needs an estimator's call. It should flag, not guess.
If most of your work is one-off bespoke fabrication with no repeatable BOM, the draft gives you less than your own estimating judgement already does.
A galvanising or finishing spec it can't tie to a real process cost is a number it should leave for a person, not invent.

The honest test

If you cannot say, right now, whether the material prices and lead times in your BOM are current this week — this tool makes your wrong quote faster, not safer.

Point it at a drawing with an unstated tolerance, or a BOM whose material prices and lead times were last touched months ago, and it drafts a clean, professional quote built on numbers that no longer hold. That is the trap.

What it doesn't do — and shouldn't

It drafts. An estimator approves. That boundary is deliberate.

WHAT IT DOES

Surfaces each extracted requirement and tags its source — email body, drawing, spec page

Shows the BOM line and lead time it priced against

Flags every requirement it could not match cleanly and every tight deadline

WHAT IT WON’T

Send the quote to the customer

Commit the price or the delivery date

Decide the margin or whether to take the job

A quote is a commercial commitment: a wrong price erodes the margin, a wrong lead time risks a delivery the workshop can't meet, and a misread material or finish spec — say a hot-dip galvanised coating to AS/NZS 4680, or a tolerance that drives the machining cost — can mean a part that fails inspection. The estimator who signs the quote carries that consequence, so they stay on the decision, not the tool.

What your data has to look like

Requirements it can extract from the pack, and a BOM with current prices and lead times to match against.

Typical readiness

across orgs we see, before the first job

A BOM with current material and process costs

Needs shaping

Current supplier lead times

Usual weak point

Machine-readable RFQ inputs

Usual weak point

A finishing and process cost reference

Usual weak point

Standard parts catalogue to match against

Usually ready

The real first job

The BOM and lead times are usually the weak point — costs held in a spreadsheet someone updates by memory, lead times that live in an estimator's head. Fixing how part costs and lead times are captured and kept current is usually the real first job — larger and more valuable than the drafting layer on top. Once the inputs are clean, every quote after that is faster and right by default.

Right fit if…

You field a steady flow of RFQs in varied formats — email, PDF drawings, spec spreadsheets

Most of your work matches a repeatable BOM and known processes

You can point to current material costs and supplier lead times today

Estimating time is mostly spent reading and re-keying before pricing starts

Walk away if…

Almost every job is one-off bespoke fabrication with no repeatable BOM

Your material prices and lead times live in someone's head or a stale spreadsheet

RFQs arrive as scanned photos or hand-marked drawings a tool can't parse

You need a tool that commits the price and the delivery date for you

Open questions

The worried-buyer questions, answered straight

It can draft a wrong price or an optimistic lead time — which is exactly why nothing is sent on its say-so. It matches each requirement to a BOM line and a supplier lead time, prices material, labour, finishing and freight, and shows its working: the source of each figure, the line it priced against, a confidence score, and a flag on any tight deadline or unmatched requirement. An estimator checks those and adjusts margin before approving. The tool surfaces the numbers; the person stands behind the commitment.

It works from inputs it can actually parse — the email body, machine-readable drawings and spec sheets, your BOM. A scanned photo of a hand-marked drawing is only as good as what can be read off it, and the tool can’t tell a misread tolerance from a real one. Getting RFQ inputs and your BOM into clean, current, structured form is usually the first piece of work — and the piece that pays off across every quote after.

No. It removes the reading and the re-keying — the half-day of pulling specs out of a pack and matching them to BOM lines — so the estimator spends their time on the judgement: the margin, whether the workshop can hit the date, whether a non-standard requirement needs a different process. The quote is still theirs. The capacity it frees goes back into the harder jobs and the customer relationship.

Current to this week. Material prices move, and supplier lead times shift with demand — a quote drafted against last quarter’s costs is confidently wrong. The honest test: do you know, today, whether the prices and lead times in your BOM are current? If they live in someone’s memory, that is the first thing to fix, because the draft inherits whatever the BOM tells it.

Customer RFQs, your BOM and your pricing are commercially sensitive. Any deployment runs against your own systems and data handling, not a shared pool — we scope where the data sits and who can see it as part of the build. The demo here runs entirely on fabricated data; AusTech Industries and Sarah Chen are not real, and the prices shown are invented.

It can extract and tag them — a 304 stainless part, a hot-dip galvanised finish to AS/NZS 4680, a hole-position tolerance — and match them to a BOM line and a finishing cost where one exists. Where a spec has no matching cost line, or a tolerance drives a machining cost it can’t infer, it flags the gap rather than inventing a number. The estimator prices the judgement call; the tool does the extraction and the matching.

What it takes to build

3–5 weeks · 4 phases

Reused from template~65%

Bespoke to this skin~35%

stack · Claude · email/PDF intake · BOM/pricing rules · review UI

What it would cost

Fixed scope, fixed price, fixed dates.

Bite-sized first piece

One contained change, low risk

Pilot build

Most builds land here

Embedded support

Scale on proof

Considering this for your workshop?

The honest place to start is a bite-sized first piece — one contained change, low risk. Tell us where the quoting hurts; we'll play it back, scope it, and show you what's possible.

Book a call How we work