Mr. Lister

Mr. Lister is your automated art to published listings pipeline. Built for POD solopraneurs, with as much as there is to do, let Mr. Lister help take the tasks off your plate. Draft, Approve, Publish.

Comment

Inspiration

Small business. Solopraneurship. Print on Demand. These three things inspire my work.

  1. We can only rely on ourselves.
  2. In business, you can only trust yourself.
  3. Print on Demand doesn't actually exist. But despite that, it is a booming market with zero commonsense solutions. I see an asymmetrical upside opportunity by attempting to provide just that.

It is to the end that I present Mr. Lister. The first in a line of POD automation products, in the Mass SKUtiny ecosystem.

A platform where one can control their entire POD business, across platforms, from one location with an AWS native, AI tool-team to handle the tasks.

What it does

Mr. Lister is an automated listing generation pipeline.

User uploads art and a fully published listing is published to your shop.

Nova/Strands/AWS handles orchestration and Gemma 3 (which outperformed the frontier models for listing generation), handles the visual review and listing generation.

Seller uploads artwork:

  1. Authenticated/private upload
  2. AWS processing
  3. Gemma visual/listing intelligence
  4. Strands orchestration
  5. Printify product creation
  6. Seller reviews/approves listing
  7. Controlled Etsy publication

Listings automated. Human approval required.

How we built it

I built this as Project Architect and Senior Project Manager. I used ChatGPT as Senior Development Partner. Work as Software Engineer. Codex as Senior Software Engineer. I used Sol Ultra and Astra Ultra.

I have a very regimented build process, honed after 18 months of progressively larger and more complicated project builds.

I start with my initial concept and flush out pros, cons, roadblocks and requirements with Chat. This develops a rigid file map and phase plan. It provides space to get to the problem I want o actually solve. Then, hone down the steps to bounded tasks that Codex and achieve. From there, it works more like a lego set, where each next step builds on top of the last. Codex stays focused and limited. I discuss and develop next steps with Chat, for context on producing the next focused build phase. I remain final word every step of the way. This is my vision and my build. I simply use an available engineering team, to complete my work. I see this no different than hiring an outsourced Engineering team or working with an Engineering team at work.

Why this approach? The tradeoff for speed, is the lack of project control. Human in the loop is fast enough and I believe produces the most optimal outcome.

Challenges we ran into

AWS permissions and IAM/root profile timeout. This is the first AWS project I have worked on and coming in cold, it was a very steep onramp. I have worked with large systems that have been built upon over 20 years, so I understand how this has gotten to where it is, but I also see why certification exists.

The biggest challenge was unwinding two phases which wound up on a parallel git branch from Main. In particular phase 7, which connected the Printify Draft and Etsy publication flows was a beast -- publication aggregates, provider credential isolation, CloudFormation, Lambda, IAM, AWS deployment safeguards, idempotency (which I had never heard of before), etc. The actual nitty gritty.

Accomplishments that we're proud of

The power of the AWS platform is truly astonishing. Regardless of outcome, that I stood up an AWS SaaS product in six weeks, by myself. This is a personal accomplishment that I will carry for some time.

I felt a bit like David Tenant's Dr. Who, when he talked about staring into the unfiltered time vortex as a child, and he was never the same after that. AWS is kind of like that for the uninitiated.

What we learned

Yet again, I learned that this is the future of software engineering. The software engineers know it, the hyperscalers know it, the builders know it. The only people who don't are the mainstream. And Human in the Loop is the only way this works. I mean, local is really the only way, but for another discussion.

I learned again, that any system can be learned. Every build, has a solution. Any idea, can be built. And opportunity is there for everyone, if they just take the time to look around. Solve something in your world that keeps having to be done the same way.

What's next for Mr. Lister

• User onboarding - user store set up form, oauth/account handling, shop handling, ability to select print provider products directly, etc • Expanded product editing - Edit image size/placement, Mockups, eventually automated product placement • Watcher - hot folder connected to Mr Lister. Art submitted there, will be picked up by Mr Lister and processed immediately. • Dashboard - view of stores, products, drafts, failures, etc.

Built With

Share this project:

Updates

Submission history