Building ShopMemo as a solo developer: giving AI a memory for ecommerce content

fenghui2 yang — Sun, 02 Aug 2026 14:25:41 +0000

I built ShopMemo because AI content tools kept forgetting the useful parts
I have tried a lot of AI writing tools while working on ecommerce ideas. Most of them can produce product descriptions, social posts, ad copy and emails in seconds.
That part already works. The part that kept bothering me was what happened after publishing.
A seller would test a hook, adjust the angle, find a better way to describe the product, learn something from the numbers, then lose that learning inside a chat history, a spreadsheet, or their own head. A month later, the next content task started from an empty prompt again.
That is why I started building ShopMemo.

The problem I saw in ecommerce content workflows
A simple ecommerce content workflow looks like this:
Find a product
↓
Create content
↓
Publish
↓
Check performance
↓
Repeat
Experienced sellers pick up a lot along the way. They learn which hooks get attention, which product angles convert, which tone fits their audience, and which formats are worth repeating.
Most of that knowledge never becomes part of the next AI run. It stays scattered across notes, screenshots, analytics pages and half-remembered decisions.
As a builder, this felt like the wrong place to stop. If AI can help create content, it should also help preserve the reasons a piece of content worked.

Most AI content workflows start too clean
The usual AI workflow looks like this:
Product information
+
Prompt
↓
Generated content
That is fine for a first draft. It is weak for a real seller who has already run campaigns, talked to customers and learned from past posts.
The AI usually does not know that last month a certain headline style worked better, that buyers in this niche respond to a specific angle, or that the brand should avoid a certain voice.
So the seller keeps teaching the same lessons again. I wanted ShopMemo to reduce that repetition.

The idea behind ShopMemo
ShopMemo tries to save the strategy behind good content. The final text is only one artifact from the work.
When a seller finds a post that worked, I do not want the tool to store it as “a good post” and move on. I want it to pull out the useful parts:
Audience:
Who was this written for?

Hook:
Why did someone stop scrolling?

Angle:
What made the product feel relevant?

Structure:
How was the message arranged?

Brand voice:
Why did it sound right for this store?

Feedback:
What happened after publishing?
Once those pieces are saved, the next content draft can start with memory instead of guesswork.

How I approached the first version
I kept the first version small on purpose. I did not want to build a general AI assistant with a vague promise. I wanted one workflow that I could test with real ecommerce content.
Analyze successful content

↓

Extract the strategy

↓

Save the pattern

↓

Generate new content using past experience

↓

Learn from feedback
The feedback loop matters most. Good content usually comes from experiments: publish, observe, adjust, repeat. ShopMemo is my attempt to make those experiments easier to reuse.
I am still early. Some parts are rough, and I am learning where memory helps and where it gets in the way. That is part of the reason I want to write about the build process in public.

What I learned while building it
AI makes production cheaper. That changes the value of experience.
Before AI, a seller might ask: can I create enough content? Now the better question is: can I create content based on what my store has already learned?
Anyone can ask a model to write another post. The advantage comes from the context around the post: customer feedback, past winners, failed angles, brand rules and small details from the business.
That context is where a personal tool can beat a generic tool.

The direction I am exploring
I think more AI apps will move from “generate this for me” to “help me work with what I already know.”
For ecommerce sellers, that means AI should remember the store, the audience, the past campaigns and the lessons from previous content. It should make the next draft feel less like a blank page.
ShopMemo is my attempt to build in that direction. I am starting with ecommerce content because the feedback loop is clear: sellers publish, the market responds, and the tool can learn from that response.

About ShopMemo
ShopMemo is an early-stage project for ecommerce sellers who want to turn their own content experience into reusable AI memory.
I am building it step by step and sharing what I learn along the way.

Website: https://shopmemo.app

DEV Community: fenghui2 yang

Building ShopMemo as a solo developer: giving AI a memory for ecommerce content