Creating quotations with AI: a standalone assistant or quotation software?

Two routes: a standalone AI assistant or quotation software

Drawing up a business quotation calls for two different skills: on the one hand putting your project approach and added value into convincing words, on the other hand drawing up a watertight financial calculation. When software is used, you see two different ways of working in practice:

• Standalone generative language models: You use a chat interface (such as an online language model) to have an introduction, problem analysis or scope description drafted from a briefing. You copy the generated paragraphs manually into a text document or layout program, after which you add the calculation table yourself.
• Specialised quotation software: Software that manages the complete route of calculation, layout, sending and digital signing. Within this category we distinguish regular quotation software and packages with integrated AI support:

Among the regular quotation tools, Offery, according to the information on the official Offery website, focuses on standardised online workflows, reusable building blocks and digital signing. In this research, no specific AI writing feature was identified for Offery; in this comparison the platform serves as a representative of structured, non-AI-driven quotation software.

By contrast there is software such as Offorte, which on the page about Offorte AI quotation software explicitly documents a built-in AI writing assistant. With it, users can have texts generated or rewritten directly within the quotation layout. In a trial, check what the assistant changes and whether amounts and terms remain unchanged when you only ask for text editing.

Practical comparison with an identical fictitious briefing

To investigate how both methods work in practice, we use a fictitious scenario in which an IT service provider submits a quotation to an accountancy firm with 12 workstations. The request concerns a migration to the cloud, security set-up and monthly workplace management.

The fictitious prompt for a standalone language model

With a standalone language model, you enter an instruction to generate a draft text:

Fictitious prompt example:
"Write a business introduction and problem analysis for a quotation to Accountantskantoor De Vries. Starting points: migration of 12 workstations to a cloud environment, implementation of two-step verification and preventive management. Use a professional and calm tone. Leave specific amounts and response times open as temporary markers such as [AMOUNT] and [RESPONSE TIME], and do not invent guarantees that are not in this briefing."

How to assess the draft

Carry out the instruction in a trial of your own. Check whether the draft mentions twelve workstations, two-step verification and management. Amounts and response times must remain open. If the assistant has nevertheless added a response time, backup frequency or guarantee, remove it and check the rest again. A convincing wording can suggest an agreement that you have not made with the client.

Note how much time you spend on checking and correcting. Compare that with adapting an existing text template. This fictitious briefing is a demonstration task; we have not assessed any model output as a test result.

Processing in specialised quotation software

In quotation software, have the same assignment built up from the available services and rates. Use, for example, twelve fictitious workstation licences of €20 per month: together €240. Add ten migration hours of €90: €900 one-off. Check whether one-off and monthly amounts remain visible separately. These are our own calculation assumptions, not a supplier's prices.

With built-in writing help, as described by Offorte, you can have text adjusted within the quotation. During a trial, check which parts change. Ask only for a shorter introduction and then compare amounts, terms and conditions with the previous version. Do not assume that a writing assistant cannot possibly affect the calculation.

Selection criteria: standalone AI chatbot versus quotation software

The table below contains selection questions. Have the supplier show the chosen version of the software; features and access rights may differ per subscription.

Three practical points of attention with quotations and AI

Anyone considering using AI when creating quotations must take three operational preconditions into account.

1. Calculation certainty versus language models

Language models are designed for producing language, not for mathematical calculations. As soon as a quotation contains volume discounts, multiple VAT rates or instalment payments, you must rely on a calculation module. So do not let an AI model calculate or add up prices on its own; use a table in which formulas or software modules manage the numbers.

2. Privacy and confidentiality of client data

When entering assignments into public, free chat assistants, there is a risk that the data entered will be used to train models further. So never enter non-anonymised client names, specific margin calculations or personal data into public tools. With business AI software, ask explicitly about use for training, retention periods and access to the input. Check the agreements; a business subscription alone does not prove that training use is excluded.

3. Version control, tracking and follow-up

A successful quotation process does not stop at writing the text. You want to know whether the client has opened the document, when it is time for a reminder and how the acceptance and the accepted version are recorded. A standalone chat without set-up integrations does not manage that follow-up, whereas specialised quotation tools send notifications and register digital signatures.

Selection guide: which way of working suits your organisation?

The optimal way of working depends on the volume of your proposals and the need for tailored texts:

• Few quotations with intensive customisation: If you write only a few large advisory proposals a month, you can perfectly well use a standalone language model as an editor for your introductions, provided you enter amounts and delivery dates manually and critically check all texts.
• Regular quotations with fixed services: look into a package such as Offery when reusable components, sending and follow-up matter more than generating new texts. Have that route tested with your own example.
• Quotation layout and writing help together: include Offorte if you want to adjust text in the same environment. Compare the correction time and the licences needed with your current way of working.

See our complete overview of quotation software to compare providers on templates, interactive elements and digital signing.