Writing quotes and proposals with AI: from request to proposal

Ask a salesperson or estimator where the time goes, and you almost always hear the same thing. Not the calculating or thinking about the price, but everything around it:

• Retyping requests. The customer emails a question, sends a PDF or fills in a form. Someone copies the details into the CRM, the ERP or a spreadsheet.
• Searching for a past quote. For a similar customer or a similar project. Somewhere in a folder, a mailbox or the archive, hoping the old price is no longer in it.
• Prices in several systems. Products in the ERP, customer agreements in the CRM, a price you still need to request from the wholesaler, a price list as a PDF.
• Standard texts by hand. Assumptions, exclusions, delivery times and terms, cut and pasted again every time.
• Waiting for that one colleague. The one who knows what something like this costs, which mark-up you charge for this kind of work and what went wrong last time. And who is busy, or on holiday.

That applies to an installer and a contractor, to a manufacturer and a wholesaler, and just as much to a consultancy writing proposals. The pattern is the same: something comes in, someone reads it, looks something up and retypes it. That's exactly the work AI can take over.

Not every quote is equally suitable. A good candidate has three things: a lot of them come in, the input is digital (an email, a form, a PDF) and you can check the outcome against a price list or past work.

• Standard quotes from a fixed range, with prices from the ERP or the price list. A logical place to start.
• Bespoke quotes with an estimate, as at an installer, contractor or manufacturer. AI prepares a first estimate based on comparable work, and the estimator checks the assumptions.
• Extra work, drawn up from the job sheets, ready for the customer's approval.
• Sales proposals and advisory proposals in professional services, with the customer's question, the approach and the right references.

Less suitable: a quote that is completely different every time, or a large project where the negotiation itself is the work. There, AI can at most summarize or make a first draft.

Not sure? Test it with the seven questions in which processes can you automate.

Less than you think, but it has to be right. AI is only as good as the information it gets.

• A few good example quotes per kind of quote. Quotes you would send again as they are, in the structure and tone you want to see.
• A price list or access to the ERP. Preferably the source itself, so AI calculates with current prices.
• Your standard texts. Assumptions, exclusions, delivery times, terms and references, in one place.
• Clear rules on margins. Which margin for which kind of work, up to what amount a salesperson decides alone, and when the estimator or management takes a look.
• Past quotes and estimates. For bespoke work: the more past estimates and post-project cost reviews there are, the better the first version gets.

The quality of that data matters. A messy CRM, last year's price list or quotes spread over ten folders: AI doesn't make that better. If it's a mess, cleaning up is step one. Missing or messy data is one of the reasons AI projects stall, see why AI projects fail.

There are roughly three ways, and they don't exclude each other. They match the four phases of AI adoption: prompting, assistants, workflows and agents.

A chat assistant with your template

A simple start. In ChatGPT, Claude or Copilot you create your own assistant with your example quotes, standard texts and instructions. A salesperson pastes in the request and gets a first version back. That's phases 1 and 2. It works without connections, but you look up and check prices and customer details yourself. How to build such an assistant is in build your own AI assistant. Writing good instructions is covered in prompting.

The quoting module in your own software

Many CRM and ERP packages have a quoting module, and some come with AI features. If the quote is created entirely in that one package, that's a logical choice. What the package can do differs per vendor. It gets difficult as soon as the information sits in several systems.

A workflow or agent connected to your systems

Here the looking up and retyping disappear too. A workflow or AI agent reads the request, looks up prices and products in the ERP, pulls the customer from the CRM, finds comparable quotes and prepares the draft. That's phase 3 and 4. For that, AI needs access to your systems, see connecting AI to your systems. We usually build this in n8n, with custom work where needed.

A logical order: an assistant with your template for everyone, and a connected workflow for the quotes that take a lot of time. Below you'll find the three side by side.

Don't start with the tool, start with the time. In our APO method, every process step gets three numbers: how long it takes, with how many people, and how often a month.

A worked example for quotes and estimates: forty minutes, two people, eighty times a month. That's 1,280 hours a year. At € 45 per hour, this process alone costs € 57,600 a year. This is a worked example, not a client figure.

AI doesn't take all of that away. We never calculate with 100 percent savings, and at most 90 percent per step. The salesperson keeps checking, the estimator keeps checking the assumptions. But if AI removes half of those forty minutes, that already saves 640 hours a year.

From practice: at one of our clients, writing a sales proposal went from 4 hours to 1 hour. Not because the salesperson sends the draft blindly, but because the first version is there, in the right structure and with the right references. The salesperson finishes it.

You can work it out yourself with the AI savings calculator. How to turn it into a full business case is in the AI business case.

We start with the AI Quick Scan: one day on site with you, with the people who make the quotes. We walk through the process from request to signed quote and give each step its three numbers. Within a week you have a business case per opportunity and a roadmap. The Quick Scan costs € 6,000.

After that we build within the AI Partnership, from € 4,500 a month, for a minimum of twelve months. The first process goes live in the first quarter, often requests into the CRM or the draft quote. Connected to your CRM and price list, tested on real quotes, with a salesperson who approves everything. We only start if the partnership pays back at least 2x in the first year. All amounts exclude VAT.

What else AI does across sales and other departments, from request to follow-up, is on AI by department.