Concept — Unshipped Design Lead AI Exploration

Millwork AI Picker

A conceptual exploration extending Lowe's AI-powered visual matching into millwork — helping customers find the right entry door through conversation or inspiration, then see it on their own home before they buy.

$AI Door Finder Assistant showing a natural-language description, a top-matched door with reasoning, and the door visualized on a home$

Before the Full Story

Arriving at a solution

The Problem

Entry doors are a high-consideration purchase — customers second-guess style, material, and how it'll actually look on their own home, the same uncertainty AI Color Match already proved it could solve for paint.

The Decision

A Millwork AI Picker reachable two ways — a natural-language request to an AI assistant, or an inspiration image upload — returning matched doors with a specific reason for each, then letting the customer visualize any option on their own home.

Where This Stands

Conceptual and unshipped — a mocked interaction, no engineering behind it, no user testing, no numbers beyond the logical case that a pattern already proven elsewhere might transfer here too.

Why Include It

Craft and judgment show up before something ships too — in how a flow gets structured, and in noticing that a pattern already proven elsewhere might transfer somewhere new.

Context

A pattern worth testing on something bigger

AI Color Match proved something on Lowe's Paint Color Wall: letting a customer visually confirm a match — not just browse a catalog of swatches — measurably increases their confidence before they buy. This concept asks the same question for a much bigger, higher-consideration purchase. Entry doors aren't an impulse buy; customers second-guess style, material, and how it'll actually look on their own home. What if finding the right one worked the same way visual color matching does?

The Idea

Two ways in, one confident answer

The concept centers on a Millwork AI Picker, reachable two ways: a natural-language request to a conversational assistant ("something modern with more glass"), or an upload of an inspiration image the customer already likes. Either path returns a set of matched door options — each with a short, specific reason it was recommended, not just a spec sheet. From there, the customer can visualize any option directly in their own home. In this flow, that means their actual front entryway, using a photo they upload themselves.

Key Moments

Where the AI actually helps.

Two paths, one confident match The full flow, end to end

01 / 02

Two paths, one confident match

A customer arrives at the Entry Door AI Picker from the product listing page. From there they can type a natural-language request directly to the AI Door Finder Assistant, or upload an inspiration image and let the AI match against it. Either path lands on the same kind of result: a short list of doors, each visualized and paired with a plain-language reason it fits what they asked for — not a filtered spec table.

Upload Inspiration Image modal, letting a customer scan a QR code or upload a photo instead of typing a description

The second path in: an inspiration photo instead of a description, scanned in from a phone or uploaded straight from desktop.

02 / 02

The full flow, end to end

The diagram below walks the whole concept: entry point, the AI assistant's chat and image-upload paths, matched results visualized against a generic scene, and the final step — uploading a photo of the customer's own entryway so the recommended doors are visualized on their actual home before they configure and buy.

$Millwork entry door AI picker flow, from entry point through AI chat and image upload paths to visualizing doors in the customer's own home$

Scroll horizontally to explore, or click to expand into a full-resolution, pannable view

Two entry paths — conversation or inspiration — arrive at the same kind of confident result.

Where This Stands

Unshipped, by design of this being early

This is conceptual, unshipped work — no build, no pilot, no numbers. Included because it's representative of how I think through applying a proven AI pattern to a new category. The flow above is the extent of it: a mocked interaction, not a working product. There's no engineering behind it yet, no user testing, no business case beyond the logical one — AI Color Match already showed this pattern works for a lower-stakes purchase, and millwork is a natural next test of whether it holds for a bigger one.

Reflection

Why show unshipped work at all

Most of what's on this site is proven — shipped, piloting, or backed by real numbers. This isn't, and I want that to be obvious rather than blurred. I'm including it because craft and judgment show up before something ships too: in how a flow gets structured, in noticing that a pattern already proven elsewhere (AI Color Match) might transfer somewhere new, in thinking through where AI genuinely helps a decision versus where it's just decoration. This is what that thinking looks like before it becomes a real project.