6:13

In this presentation Summer Chang shares the motivation, design, and implementation of dōsan (also branded as adosan.tech), an AI‑driven home‑repair assistant she built in a few days using AWS services. She explains how the app solves the common homeowner frustration of not knowing whether a problem is a quick fix or requires a professional. The user flow consists of six steps: snap a photo of the issue, answer optional quick questions, receive a diagnosis, cost estimate, part recommendations, and the option to hire a vetted local professional while staying anonymous. Behind the scenes the app runs on Amazon Aurora for the relational database, Amazon S3 for photo storage, and Amazon Bedrock for the generative AI that analyses images and text. Privacy is enforced by separating personally identifiable information into a locked table that contractors cannot access until the homeowner permits it. The business model offers the app free to homeowners, earning revenue when professionals receive qualified leads or when users purchase recommended parts. Summer invites viewers to try the service at adosan.tech and outlines a future vision where the platform proactively prevents home issues, preserving home value.

Chapters

More from Summer

0:23
youtube.com

What makes your happy? (Part 4)

2w ago
1:34
youtube.com

Figma drawing brush and motion #figma

3w ago
0:24
youtube.com

What makes you happy? (Part 2)

3w ago
2:11
youtube.com

Figma Agent Skill

3w ago
1:02
youtube.com

Stateside: OpenAI Build Week 2026

2mo ago
2:12
youtube.com

Luma's Uni-1 API

3mo ago
2:37
youtube.com

I Built a Receptionist That Calls Leads Back in 10 Seconds

3mo ago
0:54
youtube.com

Undertone: A website background generator

4mo ago
3:15
youtube.com

Undertone is an animated background generator for websites.

4mo ago
1:40
youtube.com

No photo shoot — just old renders and Luma Uni-1

4mo ago
0:46
youtube.com

Voice Agent for Websites Github Repo Template - Free

4mo ago
See all content