Skip to content
Neural Soft

Use case

AI agent for preparing quotes

An AI agent reads each quote request, checks prices, stock and lead times in your systems and prepares it for a person to review and send. Reply in minutes, not days.

  • Understands requests by email, form or WhatsApp, even when they are incomplete.
  • Uses your real price lists, customer discounts and availability.
  • Nothing goes out without a person's approval.
Joinery workshop with workbenches and tools.

What an agent is

It does not just answer: it checks, calculates and prepares

An assistant answers questions. An agent also uses tools to carry out a multi-step task, always within the limits we set.

  1. Understands the request

    Which products or services, quantities, deadline and customer.

  2. Checks your systems

    Price lists, customer terms, stock and delivery times.

  3. Calculates and drafts

    Lines, discounts, shipping, total and the email text.

  4. Asks if something is missing

    If the request is incomplete, it suggests the question to ask the customer.

  5. Waits for approval

    A person reviews, adjusts and sends.

How it works

From the request to a quote ready to send

  1. The request arrives

    Email, web form or message.

    “40 F-220 filters by the 20th”

  2. Identifies the customer

    Finds their record and terms in the CRM or ERP.

    Ruiz Garage · price list B

  3. Checks price and stock

    Uses tools connected to your systems, read-only.

    Available · delivery in 3 days

  4. Prepares the quote

    A document in your format and the reply email.

    Q-2026-311 · €780.00

  5. Human review

    A person approves

    Checks margins and terms, adjusts if needed and sends.

    Review and send

Examples

Real requests, prepared replies

Clear order

Email from a regular customer

“Hi, I need 40 F-220 oil filters and 10 sets of PF-31 brake pads by the 20th. Thanks, Ruiz Garage.”

Quote prepared

Lines
2 · price list B with 8%
Total excl. VAT
€780.00
Delivery
In 3 working days

Ready to review and send

Incomplete request

Web form

“I'd like a price for annual maintenance of the machines in my unit. We are a new company.”

Reply prepared

The agent does not make things up: it suggests asking how many machines there are, which model and where, and attaches the indicative prices.

Clarification email to review

Product out of stock

Customer message

“Can you send me a quote for 200 units of the SP-10 bracket for next week?”

Quote prepared

Stock available
120 units
Remainder
80 units in 12 days
Alternative
SP-12 available

Two options for the sales rep to choose

Who does what

What the agent does and what a person decides

The agent does

  • Reads and understands the request.
  • Checks prices, terms, stock and lead times.
  • Calculates lines, discounts and the total.
  • Drafts the quote and the email.

A person decides

  • Approving and sending every quote.
  • Special discounts and off-list margins.
  • New customers or unusual orders.
  • Any change to prices or terms.

Security

An agent with clear limits

  • Read-only

    It checks prices and stock, but cannot change prices, create orders or send anything on its own.

  • Everything is logged

    Every step the agent takes is recorded, so you know where each figure comes from.

  • Tested before use

    We evaluate it with real past requests and compare its quotes with the ones the team produced.

Technical drawings spread out on a table.

Your rules

The agent applies your rules; it does not make them up

Discounts, minimum orders, delivery charges or lead times: the agent works with the terms you already use and sets them out so anyone can check them.

  • Prices and terms come from your own systems.
  • Anything outside the rules is flagged for review.
  • Whoever approves the quote can see where each price comes from.

Method

We start with one frequent type of request

  1. Analyse the requests

    We look at how requests arrive, which data is needed and where prices and stock live.

    Map of the process and systems

  2. Build and evaluate

    We connect the agent to your data read-only and test it with real requests from the last quarter.

    Agent evaluated on real cases

  3. Go live

    We integrate it into email or the CRM with human approval and measure response time.

    Agent in use and monthly follow-up

Related service

  • AI process automation

    We automate repetitive tasks with AI so your team spends less time copying data, reviewing documents, classifying requests or drafting replies by hand.

Related training

How long does it take you to reply to a quote request?

Tell us how requests reach you, where your prices and stock live and how many quotes you prepare a month.