Why AI?

Why we decided to use AI

Writing a good quotation takes real time: describing the work properly, pricing it, wording it so it reads professionally. Most trade business owners we spoke to do this at 9pm at a kitchen table, not because they enjoy it, but because there was no faster way to get it right. AI turned out to be genuinely good at exactly that specific task: turning a rough description into a properly structured first draft. That's the whole reason it's here: not because AI is fashionable, but because it solves a real, specific, time-consuming problem.

What AI is actually good at, here

• Turning a messy, spoken description into properly structured, well-worded text.
• Matching a tone or style once it's seen a few real examples of your own writing.
• Sorting a job into the right category from a list you've already defined.

What it isn't good at, and shouldn't try to be

• Measuring anything from a photo, or inspecting a property it hasn't seen.
• Knowing your prices, unless you tell it what you charge.
• Judging a specific customer relationship: whether to give someone a discount, how to handle an awkward conversation, anything that needs actually knowing the person.
• Being right every time. It won't be. That's exactly why review comes before send, not after.

Why every suggestion is reviewed by a human

Because a quote sent with a wrong assumption in it costs you a customer's trust, not just your own time. Every draft (a quotation, a categorised job, a business overview during setup) stops in front of you before it goes anywhere. That's not a safety net we added afterward. It's the actual design: the software drafts, you decide.

How it learns each business

Not by watching you work. By what you deliberately teach it: a short written overview of your business, examples of quotes you were proud of, corrections when it categorises a job wrong. A business that's taught it nothing still gets a competent draft; a business that's taught it a little gets one that sounds more like them, specifically.

Why we never let it silently make important decisions

Because “silently” is where trust breaks. A price that changes without you seeing it, a message sent without your review, a category assigned without a way to correct it: any of those would make the software feel like it was working against you instead of for you. So none of them happen. If AI touches something that matters, you see it first, every time, no exceptions.

Company Knowledge and Favourite Quotations

Two plain-text settings that do the actual teaching: a short description of what your business does and how you like to work, and up to five past quotes you mark as genuinely good examples of your own voice. Neither is required. Both are entirely optional, editable, and removable at any time.

How your data is actually used

When you generate a quote, the description and any photos you attach are sent to OpenAI's systems to produce the draft, the same as if you'd typed them into ChatGPT directly. That request is governed by OpenAI's standard API terms, which don't use API data to train their models by default. TradeFlow365 doesn't share what you write with other companies using the platform, and nothing about your business is used to change what another company sees. For the full legal detail, see our Privacy Policy.

Why TradeFlow365 improves over time

Every Favourite Quotation you save, every category correction you make, every line you add to Company Knowledge sharpens what the assistant produces for your business specifically. It isn't a shared model that changes for everyone based on everyone else's data. What gets better is your own results, because of what you specifically taught it.