RFQ to quote automation turns an emailed request for quotation into a priced draft quote in your ERP. 3 kinds of tool do it: CPQ modules and ERP assistants that expect a clean request, inbox parsers that extract line items, and AI agents that match every line to an article, price it from the ERP and hand a salesperson the draft.
At Lleverage we think most buyers judge these tools on the wrong step. Producing the quote document is easy: every ERP can print one. The hours go into reading the request and working out what each line means when it has no article number, carries another supplier's code or says "same as last time". Judge a tool on how it handles those lines in your own requests.
This guide follows one emailed RFQ from the inbox to a draft quote a salesperson can send. It covers what the request looks like, how matching works, where prices and stock should come from, what a person checks and how to compare the options. Our quote and order agents do this work inside the ERP you already run. If you would rather see it on your own requests, book a demo and bring 10 of them.
Which solutions can turn emailed requests for quotation into draft quotes?
3 kinds of software can produce a draft quote from an emailed request. They differ in how much of the reading they take on. A CPQ module needs a person to read the request and type it in, and an ERP vendor's assistant reads it within the limits of that ERP. Inbox parsers read the text and leave the matching to you. AI agents read, match, price and draft, then stop for a person.
| Type of tool | What it does with an emailed RFQ | Where it stops | Fits best when |
|---|---|---|---|
| CPQ module, ERP quote screen or ERP assistant | Builds and prices the quote from a clean request; a vendor's assistant also reads the email and drafts a quote for review | Needs a structured request, or works inside one ERP with the matching that vendor built | Products are configurable, or you run that ERP and your requests are fairly clean |
| Inbox parser or RFQ extraction tool | Pulls line items, quantities and dates out of the email and attachments | Leaves article matching and pricing to a person or a rule table | Requests follow a few fixed layouts with your article numbers |
| AI agent working in your ERP | Reads the request, matches each line to an article, prices it from the ERP, drafts the quote and flags unclear lines | Hands the draft and the flagged lines to a salesperson to approve | Requests arrive in many formats, with vague or foreign codes |
A manufacturer of configurable machines still needs configuration rules, and our post on CPQ for manufacturing covers that case. This guide is about the other case: a wholesaler or a manufacturer of standard parts that gets dozens of line-item requests a day by email, each one with 5 to 200 lines to price.
Where an ERP vendor's own assistant fits
ERP vendors now ship their own assistants for this work. Microsoft's Sales Order Agent for Business Central is the clearest example. It watches a mailbox and reads the email with its PDF, PNG and JPG attachments. It finds the customer by the sender's email address, looks up the items and their availability and creates a sales quote. It always creates a quote first, even when the customer asks for an order. A Business Central user reviews every outgoing email before it is sent.
Its item search tries the item number, the vendor item number, the GTIN and the item references first, then item attributes, then a broader search across all items. Microsoft notes that the quality of your item descriptions, attributes and categories decides how well it finds products. It runs on Copilot Credits. A forwarded email is matched to the person who forwarded it, so someone links the right customer by hand. For a Business Central company with tidy item data and customers who email directly, it is worth a test. Run it on your messiest requests as well as your clean ones.
What does a typical emailed RFQ look like in wholesale and manufacturing?
A typical emailed RFQ is written for a person. It mixes a few lines in the email body with an attachment in the customer's own format. It uses the customer's words for your products and leaves out details the sender assumes you know. Very few arrive as a clean list of your article numbers with quantities.
On a normal morning, a sales desk at a technical wholesaler sees requests like these:
- Free text in the email body. "Can you quote 40 of the stainless hinges we had in spring, plus 2 boxes of the M8 bolts, delivered to the Tilburg site."
- A PDF from a purchasing system. A formal RFQ with the customer's own item codes, a requested delivery date and terms on page 3.
- An Excel sheet in the customer's layout. 120 lines, columns named in Dutch or German, units in pieces on some lines and in boxes on others.
- A scanned or photographed list. A page from a maintenance engineer's notebook, or a photo of a worn part with "need 6 of these".
- A forwarded chain. The real request sits 3 emails down, below a reply from someone in the customer's purchasing team.
Many of the lines inside these requests share the same problems. Some have only a description ("ball bearing 6204 sealed"). Some carry the customer's own code or another supplier's catalogue number. Some point back in time ("same as order 2291", "as last year"). Some mix a standard item with a question about an alternative.
At SIG Benelux, a building materials distributor, around 3% of order lines carried an article number and roughly 47% had only a description. Those were orders, one step after the quote, and the quote requests we see from builders' merchants and technical wholesalers are written the same way.
What are the steps from inbox to draft quote?
An emailed RFQ becomes a draft quote in 7 steps. The agent sorts the email, reads it, identifies the customer, matches every line to an article, prices each line and checks stock. Then it creates the draft quote in the ERP and hands it to a salesperson. Steps 4 and 5 hold most of the work.
- Pick up and sort. The agent watches the shared mailbox and separates quote requests from orders, questions, complaints and newsletters. A request for quotation is routed to the quoting flow, and anything else goes where it belongs.
- Read the request. It reads the email body and the attachments: PDF, Excel, CSV or a photo. It pulls out each line with its description, code, quantity and unit, and the requested delivery date, delivery address and the customer's reference.
- Identify the customer. It matches the sender to a customer in the ERP. When the email was forwarded or the sender is new, it asks a person to confirm which customer the request belongs to.
- Match each line to an article. It finds the article number for every line, using your item data and the customer's history. This step takes the most judgement, and the next section covers it.
- Price each line. It takes the price from the ERP for this customer: the price list, the customer's agreement, volume breaks and any project price.
- Check stock and lead time. It looks up availability and the lead time for the requested quantity and date, and notes a line it cannot deliver on time.
- Create the draft and hand it over. It creates the sales quote in the ERP and sends the salesperson a short summary: what it matched, what it priced, and which lines it was unsure about, with the reason.
A tool that skips steps 3 and 4 produces a nicely formatted quote with the wrong items in it. A tool that does all 7 still stops before sending, because a quote is a commitment and a salesperson owns it.
Our agent creates the sales quote in the ERP itself. In Business Central it creates and looks up sales quotes, in SAP S/4HANA and Dynamics 365 Finance and Operations it creates and looks up sales quotations. The quote sits in the ERP with your own numbering, terms and layout.
How does article matching work when the customer uses its own or a competitor's codes?
Article matching links each requested line to one article in your catalogue. It works in layers: an exact code match first, then a match against the customer's own codes and past orders, then a match on description and attributes. When none of those gives one clear answer, the line is flagged for a person with the likely candidates listed.
The layers of a good matcher
- Your own article number. This is the easy case and the rarest one. A check digit or a typo still needs catching.
- The customer's own code. Many ERPs can store a customer's item code against your article. Business Central calls these item references. When the reference exists, the match is exact.
- Another supplier's code. A customer switching to you sends a list with a competitor's catalogue numbers. A good matcher uses the cross-references you hold, the manufacturer's part number and the attributes behind the code (size, material, standard) to propose your equivalent.
- The customer's history. "The stainless hinges we had in spring" means something specific to this customer. Their past quotes and orders say which article that was.
- Description and attributes. "Ball bearing 6204 sealed" resolves to one article when your item data carries the type, the seal and the size.
In our experience the history layer does the most work, and a rule table cannot hold it. Years of orders already hold the translations your desk made by hand: this description, from this customer, meant that article. At SIG Benelux the matching uses that order history in place of a rulebook, and 88% of order lines now match the correct article on the first try, including lines that arrive without an article number.
"I knew the technology made this possible. My biggest question was the quality of our data. It was a positive surprise to see the order history was enough of a base to get results this fast." Ditte Smit, Marketing Director, SIG Benelux
What the matcher should do when it is unsure
A matcher that is unsure should say so. The worst outcome on a quote is a confident wrong match: the customer receives a price for the wrong bearing, accepts it, and the mistake surfaces at goods receipt. A good tool shows the salesperson the line, the 2 or 3 candidates it considered and why it could not choose.
Each answer the salesperson gives should improve the next match. When they pick the right article for a competitor's code, a good tool keeps that link. The same code from the same customer then resolves on its own. In Lleverage this sits in Skills: the agent proposes a change to how it does the job from the feedback it gets, and a person approves the change before it goes live.
Where should prices and stock in a draft quote come from?
Prices and stock in a draft quote should come straight from the ERP at the moment the draft is made. That means the customer's price agreement, the price list, the volume breaks and the current availability. A tool that keeps its own copy of the price list can quote an outdated price.
Check this before you buy. A tool that matches against a price file exported once produces its quote outside the ERP. That quote has to be retyped into the ERP when the customer says yes, and the price in it may already be out of date.
A draft quote built in the ERP avoids both problems:
- The ERP knows the agreement for this customer, this item group and this quantity. The draft uses it, and the salesperson sees it is the agreed price.
- A request for 400 pieces gets the 400-piece price, and a project request gets the project price if one exists.
- Availability at the requested date decides whether the line is quoted from stock or with a lead time. A line the warehouse cannot deliver in time is flagged, so the salesperson can offer a date or an alternative.
- A price below your margin floor, or a discount above a salesperson's authority, waits for the person who can approve it.
The approval of an unusual price does not need a separate tool. Our agent can hand a colleague a task, such as a sales manager approving a discount, and carries on with the quote once the answer comes back.
What does the salesperson check before the quote is sent?
Before a draft quote goes to the customer, the salesperson checks the flagged lines, the prices that fall outside the agreement, the delivery dates and the overall sense of the quote. Clean lines need a glance. The review should take minutes, because the draft arrives with the questions already sorted to the top.
What a good draft shows the salesperson:
- The flagged lines first, each with its reason: no match, 2 possible articles, a unit that does not fit, a quantity far above this customer's normal order.
- The source next to the field. In Lleverage, the review screen shows the request's PDF beside the draft. Hover over a field and the matching value lights up in the document.
- Any price that differs from the customer's agreement, with the reason it differs.
- Dates the warehouse cannot meet, with the earliest date it can.
- Anything the customer asked that is not a line, such as a question about an alternative, a request for a certificate or a note about packaging.
Who decides what goes through on its own
At first, every draft goes to a person, so your team sees where the agent is reliable. Over time, the agent scores how sure it is of each piece of work. Below the confidence line a person decides, and above it the work goes through on its own. Only you move that line, and you can set it per process.
For quotes we advise keeping the final send with a person for a long time, because a quote commits you to a price and a date. The time saved comes from the reading, matching and pricing, and a salesperson who approves a correct draft in a few minutes keeps that time.
How do you compare RFQ to quote automation tools?
Compare RFQ to quote automation tools on your own requests. Take 30 recent RFQs, including the messy ones, and run them through each tool. Measure how many lines it matches correctly, how many it flags, and how many it gets wrong without flagging. Then check how the tool connects to your ERP and how the approval step works.
| What to test | Why it decides the outcome | How to test it |
|---|---|---|
| Line matching accuracy on your RFQs | Matching is where the time goes, and a wrong match costs more than a flagged one | Count correct, flagged and silently wrong lines on 30 of your own requests |
| Handling of vague lines | Lines with no article number, foreign codes or "same as last time" are the daily reality | Include requests with competitor codes and references to past orders |
| Attachments | Requests arrive as PDF, Excel, scans and photos, often several per email | Send emails with 2 or more attachments, one of them a scan |
| ERP connection | A quote outside the ERP has to be retyped and loses the time it saved | Ask where the quote is created and whether prices are read live |
| Approval step | A person owns the quote, and the review has to be quick | Watch a salesperson review a draft with 3 flagged lines |
| Learning from corrections | Corrections should make the next quote better | Correct a match, send the same request again and compare |
| Customer identification | Forwarded emails and new senders are common | Forward a request from a colleague's mailbox |
The number that matters most is the count of lines the tool got wrong without flagging them. A tool that flags 20% of lines and gets the rest right saves your desk time from the first week. A tool that flags nothing and gets 5% wrong sends wrong prices to customers.
In a first meeting, ask what happens to a line with only a description and no code, and whether the tool reads the price from your ERP when it drafts, or from a file it was given.
If you are also looking at the order side of the same mailbox, our checklist on order entry software for wholesalers covers the tests for orders. The two jobs share an inbox and most of the matching, so one agent can handle both.
How long does it take to automate RFQ to quote?
A first RFQ to quote flow typically runs supervised within a few weeks. Our forward deployed engineers build it with your sales desk: they read along in the real mailbox, connect the ERP, set up matching on your item data and history, and run the flow next to the team before any draft reaches a customer.
A rollout usually goes through 4 stages:
- Pick one stream. One mailbox, one customer group or one product range. The highest volume with the most retyping is usually the best start.
- Shadow run. The agent drafts quotes alongside the desk. Your team sees where it is right, where it flags and where it would have been wrong, without a customer seeing anything.
- Supervised drafts. Salespeople start from the agent's drafts and approve every quote. Corrections improve the matching as they go.
- Widen and raise the line. Add more mailboxes and customers, and raise the confidence line once the process has earned it.
Many teams worry that their item data is not clean enough. In our experience the order and quote history in the ERP carries more of the matching than the item descriptions do, and the gaps it shows are worth fixing anyway. If your item master needs work first, our master data management agents keep it in sync across systems, which helps every process that reads it.
Frequently asked questions
What is RFQ to quote automation?
RFQ to quote automation is software that turns an incoming request for quotation into a draft quote without someone retyping it. It reads the email and attachments, identifies the customer, matches each line to an article, prices it from the ERP and creates the quote there. A salesperson reviews the draft and sends it to the customer.
Can AI create a quote from an Excel or PDF attachment?
Yes. An AI agent reads PDF, Excel and CSV files and photos, in the customer's own layout, and pulls out each line with its code, description, quantity and unit. The hard part comes next: matching each line to the right article. A good tool flags the lines it cannot match with confidence, with the likely candidates.
How does the software handle a request without article numbers?
It matches the line on what it does have: the customer's own code, a competitor's part number, the description and attributes, and the customer's past quotes and orders. When one article stands out, it uses it. When 2 or more fit, it flags the line and shows a salesperson the candidates to choose from.
Does the quote get created in our ERP?
With an agent that works in your ERP, yes. Lleverage creates sales quotes in Business Central and sales quotations in SAP S/4HANA and Dynamics 365 Finance and Operations. The quote sits next to your other quotes, with your numbering, terms and layout, so nobody retypes it when the customer says yes.
Does a person still approve every quote?
At the start, yes, and we advise keeping a person on every send for a long time, because a quote commits you to a price and a date. The agent scores how sure it is of each draft. Below the confidence line a person decides. Only you move that line, and you set it per process.
How long does it take to go live?
A first flow typically runs supervised within a few weeks. Our forward deployed engineers build it with your sales desk, on one mailbox or one customer group, and run it next to the team before customers see a draft. From there you widen the scope as the matching proves itself on your requests.
How does Lleverage turn emailed RFQs into draft quotes?
Lleverage connects to your shared mailbox and to the ERP you run. The agent reads each request and its attachments, identifies the customer, matches the lines using your item data and order history, prices them from the ERP and creates the sales quote there. A salesperson reviews the draft and the flagged lines, and sends it.
A few things set the way we work apart from an inbox parser:
- It works in your ERP. The agent fills the same fields a person fills, through the system's own interfaces, so the ERP checks the work the same way. Nothing is migrated. Our ERP integrations list the systems we work in, including SAP, Business Central, Dynamics 365 Finance and Operations, Exact, AFAS and Infor.
- It reads what customers send. Email text, PDF, Excel and CSV files and photos, in the customer's own layout.
- It asks when it is unsure. An unclear line becomes a question for the right person, with the candidates and the source next to it. The draft carries on once it has the answer.
- It gets better from use. Corrections feed the agent's Skills, and a person approves each change before it goes live.
- You stay in control. Roles and scopes per connection decide what the agent may see and change, and a full audit trail records what it did, on which record and why. Our control layer covers this in more detail.
The same agent can handle the orders that follow the quotes. When a customer accepts, the order arrives in the same mailbox and the agent creates the sales order from it. Our post on sales order automation covers that step.
To start, collect 30 recent quote requests from your shared mailbox, including the ones with no article numbers, competitor codes and scanned lists. Count how long your desk spends on each, and which lines took the longest to match. That list is your test set for any tool.
Then book a demo and bring it. We will run your requests through the agent, show you the draft quotes in your ERP and walk you through the lines it flagged and why.
