Configurator AIs

AI configurator for quotes, builds and pricing, through conversation instead of a form.

Configurators have always been a hassle. Someone wants to put something together, a materials list for a project, a holiday, a catering order, and the staff member faces a form full of dropdowns, a price list, and a head full of rules only experienced colleagues know. "You'll need a concrete base for that." "That's not an option in this season." "Customer is above the threshold, so 5% discount." It feels like this doesn't belong in a form. This should be a conversation.

Here you see that conversation. The user describes what they want, often messy, with adjustments along the way, in everyday words. The AI builds the quote step by step, with the domain rules as a built-in colleague. No dropdowns. No searching the price list. No "oh, but it doesn't work that way." Just a quote that's right, and a user who actually understands it.

Hardware store

Someone wants to build a 20-meter fence. The assistant works out what they need, thinks about the concrete base, and drops the list into the cart.

Someone walks into the hardware store with one goal: a fence of twenty meters, for the garden. What he doesn't know, or forgets, is everything that comes with it. How many posts. What size. Concrete bases or in the ground. The right screws. Whether he needs stain. An experienced staff member has this list in under a minute. But that staff member isn't always there, or is busy with three other customers.

Here you see a conversation with an assistant that does have that experience. The customer says what they want. The assistant probes, what's the height, hardwood or pine, concrete base or not, and assembles the entire list. Including the screws people easily forget, and the tip that if the fence is taller than 1.80m, most municipalities require a permit. The customer goes home with materials that fit, not materials that "more or less" fit.

It's not the technology that's remarkable. It's the recognition that a DIYer doesn't need to know the entire hardware store catalog. He just needs to describe his project.

Travel agency

Building a holiday through conversation instead of filters: budget, group, preferences, constraints, the trip builds itself as you talk.

A travel agency website is usually a maze of filters. Destination, dates, number of people, budget range, all-inclusive yes/no, stars, distance to the beach, dropdown after dropdown. For people who know exactly what they want, that works. For most, it doesn't. "We want somewhere warm, not too crowded, with things for the kids to do, and no more than two thousand euros." Try translating that into six filters.

Here the assistant does that work. It probes, when roughly, how old are the kids, can it be a long flight, and builds a shortlist as you go. Not one answer, but a few options with reasoning. The customer can react to the options themselves ("the one in Italy, but a week earlier"), and the assistant adjusts.

What this adds beyond a filter: the conversation automatically does the work that experienced travel advisors do, pulling out the real question, discovering constraints the customer doesn't mention, and giving back realistic options. No overwhelming result of 200 trips.

Meals / catering

Catering order via conversation: number of people, dietary needs, budget, allergens. The order rolls out with realistic portions and the standard math.

Ordering catering through a form is a stumbling block. How many sandwiches per person? And if three of the twenty are vegetarian, how do you split it? What do you do with the two people with a nut allergy? And does it fit the budget? For the caterer, this is daily math. For the customer, it isn't.

In the demo, the customer describes the event the way they'd tell it: "twenty people, lunch, two vegetarians, one with a nut allergy, budget around three hundred." The assistant assembles the order with the portion choices an experienced caterer makes by default. And if something doesn't work, say, the nut allergy and a specific sandwich, it flags that cleanly so the customer can decide.

The pattern: software shouldn't make the customer calculate what a professional does automatically.

Demos run on request, I'll happily open one for a serious case. Ping me to walk through it together, and you'll see it work directly.