Kai Takeuchi
← Back to Everyday life

Everyday life · Independent build

Next Chapter

A resale assistant for my wife’s aunt’s senior downsizing business. Photos and voice notes become organized inventory, researched prices, and reviewed listing drafts.

My role
Product design & development
Designed for
Small senior downsizing and resale businesses
Current status
Private app
Next Chapter: from intake to a new home A couch connects six illustrated steps: a photo and voice waveform, identification with a magnifier, price comparisons, reviewed listing cards, a sale and other listings closing, then a parcel or pickup truck. Listing publication is assisted. Automatic sale monitoring and delisting are planned. NEXT CHAPTER From intake to a new home. One item record. Every step connected. 01 Capture Photos + voice notes 02 Couch Understand Parse fields + identify 03 $ Price Research + compare 04 Prepare listings Channel, copy + review 05 SOLD Sell + sync Monitor sales + delist 06 Fulfill Shipping or pickup
The end-to-end workflow. Listing publication is assisted; automatic sale monitoring and delisting are planned.

Why I built it

My wife’s aunt runs a senior downsizing business. Helping someone move also means finding new homes for their belongings: furniture, appliances, collectibles, and objects that nobody can quite identify.

Selling each item meant repeating the same manual work: take photos, figure out what it is, research a price, write a description, and post it on a marketplace. After a sale, she still had to find the physical item, arrange pickup or shipping, and remove any other listings.

I wanted to connect those tasks. Next Chapter is an orchestration layer around one item record, carrying its photos, notes, research, location, and selling history through the process.

The experience

Capture starts at a client’s house. Choose the client once, then add one item or a batch of photos. Speak a note while looking at the objects. The app transcribes it and proposes which details belong to which items and fields, ready for review.

An item can start with just a photo. It gets a short permanent code that can go on a handwritten label, plus a location so it can be found later. AI research can suggest a name, description, price range, asking price, and suitable marketplace. A seller or client’s estimate stays separate from the researched value.

The aim is a natural next step: “This looks like an Instant Pot Duo. Here is the evidence, a suggested description, and a starting price for a quicker sale.” Uncertain details stay visible and editable.

How it works

One item record, from intake to handoff
  1. Capture

    Photos and voice notes inherit the client visit. An item code and storage location connect the digital record to the physical object.

  2. Understand

    Transcription becomes reviewable fields. Images and notes give the research process a starting point, even without an item name.

  3. Price

    Research proposes an identity, comparisons, a value range, and an asking price, with sources and assumptions for the seller to review.

  4. Prepare listings

    Choose a channel and review the copy and photos. Etsy receives an unpublished draft; a Chrome helper is built to prepare Facebook’s form for seller review and publication.

  5. Sell and sync

    Record listing links and sales. Order review and removal tasks are built; live sale monitoring and automatic removal from other marketplaces remain planned.

  6. Fulfill

    Use the item’s code and location to retrieve it, then follow pickup or packing tasks. Shipping-label purchases are a future integration.

This shows the intended complete loop. Today, the seller still confirms marketplace actions and tracks the handoff; unattended selling is not enabled.

Facebook is the local-pickup path for bulky items. Etsy is an option for eligible vintage pieces. I also want to support eBay, particularly for shippable items, but I’m currently waiting for eBay developer approval.

Stack and choices

WorkspaceReact · TypeScript · Vite
Bring capture, inventory, research, and selling tasks into one interface.
Records and mediaHono · Cloudflare Workers · D1 · R2
Keep item and task state in a database, with private media storage and scheduled backups.
Research and processingOpenAI · Cloudflare Workflows
Transcribe notes, research images and web sources, and retain progress for work that outlasts a page visit.
Marketplace handoffEtsy API · Chrome Manifest V3 · eBay adapters
Prepare provider-specific drafts and connect the app to a browser helper. Activate each integration after account setup and live checks.

Decisions and design lessons

The item record is the center. A marketplace listing is one representation of an item. Keeping ownership, location, photos, and sale tasks together makes it possible to coordinate several channels without losing track of the object.

Capture should not wait for certainty. A photo and a spoken observation can be enough to begin. Identification and pricing become reviewable proposals, rather than prerequisites for saving an item.

Preparing a listing is different from publishing it. The browser helper transfers reviewed information; the seller remains in control of publication. Sale detection and delisting need confirmed provider responses before the app can call those tasks complete.

Current state and next step

The private app supports capture, structured notes, inventory, research, listing preparation, and fulfillment tasks. The Etsy unpublished-draft transfer and Facebook Chrome helper are built, with seller-account and live-form validation still to complete. Etsy developer access is approved; eBay developer approval is pending.

The next step is to validate each marketplace handoff with the seller’s account, then enable and test live order syncing and cross-marketplace delisting. The full loop is the direction of the project, with each connection added and verified in stages.

The public repository documents the implementation, setup, and integration boundaries.