The TLDR

Highlight: This week, we’re building a small custom application to handle a simple task without writing a single line of code. We chose to build a QR code!

Musing of the Week: We’re thinking about the latest “productivity boosts” we’re seeing from Google’s Duet AI. Not all of it feels healthy, and Google Meet’s new AI assistant almost seems like it was purpose-built to help people opt out of important commitments.

🔍 Idea To Micro-App in 45 Minutes

There's been a lot of buzz about GPT-4's coding skills, and we decided it was time to see if we (two non-coders with humanities backgrounds) could use take advantage as well. So, we started from scratch with the goal of creating a micro-app without getting caught up in complex coding. We decided to build a QR code, and in less than 45 minutes, we had a functional micro-app that could be distributed to a team or added to our personal workflow.

We’re excited to share some tips to get you started, building tools to solve your unique problems.

Step 1: Setting Up An Environment

GPT-4 allows you to write and execute a substantial amount of code in its internal Python environment. But, when it's time to package those micro-apps on your local device, you'll need a reliable Integrated Development Environment (IDE) and Python installed on your system. If you’re exploring coding, setting up an IDE is a step you can't skip, so this is a great place to start. Plus you’ll only have to do this once and then you can skip this step for any future projects.

Quick Guides to Setting Up Your IDE:

• Visual Studio - A free multi-language coding environment built by

  Microsoft

• Pycharm - A freemium IDE with lots of integrations and a 30-day free trial

Don’t hesitate to ask ChatGPT questions. It was able to troubleshoot our installation process quickly and easily.

Step 2: Defining Your Scope

Our next step was to lay out a goal for our micro-app. ChatGPT can absolutely be used to build large-scale applications, but we wanted to keep it quick and easy, so we built a simple tool to generate custom QR codes containing branding or images. You can copy and modify our prompt to get started on your own projects:

[Prompt]
I am building a self-contained desktop application that can be used to generate QR codes containing a logo or image. You will act as an expert python developer who is helping me to build my application. Assume that I know nothing about coding. My application should have these features:
- A simple but modern user interface
- The ability to select an image from my device
- The ability to select an output folder for the QR code
- The ability to paste a target URL
- Some styling options to control the appearance of the QR code including the resolution
The final output should be:
A high quality QR code containing the uploaded image within a small circle or square at the center of the code (user's should have the ability to choose between the two options).

This prompt launched us in the right direction. From there, it was just a matter of continuing the conversation, copying the code, and telling GPT-4 in simple language if anything was going wrong. We specifically avoided touching any of the code manually to make sure that everything we did was achievable with little to no software engineering background.

Step 3: Packaging your micro-app

Once the code was working in our IDE the last step was to package it for use as a desktop app. We gave ChatGPT the name of our file, and it gave us the exact commands to input into the IDE terminal to save the script as a functioning micro-app. End-to-end, the whole process took us less than 45 minutes, and we didn’t have to touch a single line of code!

🧠 Musing of the Week

For the past few months, Google has been teasing Duet AI, a set of AI augmentations for its Workspace suite (think: AI features for Gmail, Docs, Slides, Sheets). You may have been offered a beta version of these tools, such as the “Help me write" feature in Google Docs or contextual suggestions for your Gmail replies. The support for writing and building slides is pretty standard fare.

However, Google Meet (video conferencing tool) is introducing an AI that can join meetings with you (and for you), suggesting that you may never have to take notes, pay attention, or show up to meetings again. We have yet to see this AI in action, but it’s already got us pausing to think about whether this development is beneficial to our teams and cultures. Equipping every person with a non-contributing AI that can listen, take notes, and report back (essentially an assistant) sends a message that you don’t want people to value their team’s time or actively participate.

The Verge reports that “if everyone in the meeting sends their AI assistant, Meet will quickly figure it out and end the call.”

We like to say that AI is a tool, like a hammer, that can be used for good or bad. But this particular feature feels like it was already designed to eliminate that choice, and the innate purpose leads people to make the wrong choices (i.e., skip out on commitments and de-value your fellow co-workers’ time, energy, and thoughts).

What kind of future are we building for? What is the point of productivity culture and when does “saving time” stop mattering?

🙌 If you’re hyped about the generative AI industry specifically, here are some of the coolest roles we’ve seen this week:

• ML Content Marketing Analyst @ Clarifai

• Sr. Product Marketing Manager, AI @ Docusign

• Product Delivery Manager @ Eightfold AI

• Product Technical Program Manager, AI @ Meta

🔨 Check out these other AI tools we’ve been playing with this week:

• Jigso: AI sidekick that can be added to Slack to alert you proactively, flag questions and action items, and field your questions

• EasyListing: Generate SEO-optimized real estate listing descriptions

• Unread AI: Reads and compiles your emails, news, chats, and pods

That’s all for this week. See you next Tuesday!

Lorel & Reily