This is NOT another Cyber Monday promo

You've probably received TONS of Black Friday, Black Week and Cyber Monday promotions.

I don’t have anything special to sell so I won’t bother you with another promo (except if you want to test my AI-coded AI Jingle Maker, then drop me a line and I’ll grant you a special discount).

As far as I’m concerned, I just replied to 2 “offers you can’t miss”:

1° I purchased the Envato Prompter, mainly to improve eye contact during videoconferences. The teleprompter acts as another screen, so you can display your Google / Microsoft / Zoom call just in front of your eyes.

2° I upgraded to an annual plan on Suno, which I use on a daily basis, both for work and for pleasure. I’ve just released a brand new song on Soundcloud (I wrote the lyrics, an anonymous male singer offered me an outstanding performance).

Since the last edition of this newsletter, I’ve been AI-coding A LOT, for my own projects and for clients.

In this email, I’d like to share with you a few of my recent use cases, as well as some tips to optimize your AI-assisted coding efforts.

The Rise of Personal Software

The #1 use case for AI-assisted coding is definitely Personal Software development.

Here’s a Linkedin post from Replit’s CEO (of course, he has a vested interested in the space), shared on December 1st.

I would argue that in many cases, prompting an AI agent to craft a quick personal application is now faster than using no-code tools like Zapier or Make to connect multiple SaaS applications. It also gives you granular control on your business logic.

Here are some of the projects I’ve shipped over the last 6 weeks, as part of my business consulting activities (Remember: I’m NOT a developer).

Franchise Leads Reporting App

I’ve developed a reporting app which will be used by the Senior Management of a wellness franchise network.

The app enables franchisees and their marketing agencies to report their ad spend and lead gen results on a weekly basis, in order to compare the Cost Per Lead across branches. The app was developed and deployed with Replit AI Agent. It took me 4 hours to ship and fine tune the MVP.

Since the company manages two different networks, with a slightly different product offering, I shipped two different apps sharing the same underlying business logic.

Multilingual WordPress Plugin

For another client, I needed a multilingual content gating plugin for a WordPress site. We wanted to implement a very specific outreach workflow to potential readers (pre-vetting their email addresses to grant them free access) and make sure the UI elements and email notifications were adapted to all onboarding scenarios and languages.

I had never developed any WordPress plugin. It took me roughly one day to ship the MVP and another 7 hours to fine tune the product.

It’s now in production on 2 multilingual content sites.

Competitive Analysis Tool

Another client, a French Search Fund, asked me to develop a competitive analysis tool for lead generation purposes.

The tool enables prospects to simply enter the domain URL of their business, confirm their activity (based on a suggestion from OpenAI after scraping the content of their site) and then receive a PDF report incl. some SEO KPIs of their closest organic competitors.

The tool also creates a detailed responsive table with multiple data points, website screenshots and other enriched info, leveraging multiple APIs (ValueSERP, DataForSEO, Apollo, URLBox,..).

It took me 2 days to ship the first iteration and another 3 days to fine tune the tool based on my client’s requests.

The fund manager can now integrate the tool in his personal stack.

Lead Moderation Tool

I’m in charge of lead sourcing for a boutique advisory consultancy.

I mainly source those B2B leads through Linkedin Sales Navigator, then enrich them with Evaboot. The leads are imported into a Google Sheet for further enrichment, via a series of AI-coded Python scripts.

At the end of the process, I wanted to offer to my client a user-friendly interface to moderate the list of suggested leads (which isn’t that easy to do in a tabular presentation) before I can inject them in La Growth Machine where I orchestrate my multimodal outreach campaigns.

I used Replit AI Agent to create a custom lead moderation tool, which can be used with any authorized Google Sheet.

The client simply maps the fields from the Google Sheet (see above). He can then navigate with < > arrows through the sheet and validate the profiles via 2 buttons, NO or YES, which record the choice in the MODERATION column of the sheet.

The app can now be used for other Lead Gen projects. I also consider releasing it as a micro SaaS.

One (Simple) Stack Approach

My basic brief to the AI is always the same: I want a simple Flask web app, with a main.py file for the core logic and secondary Python files for the dedicated functions (Scraping, AI Writing, Calls to a specific API,..).

I have one main index.html in the templates folder, one main .js and .css file in the static folder.

If I need a front and an admin page, I simply create multiple .html files + one .js and one .css file per page. I can also use the base / extend logic in my html templates when it makes sense.

I don’t use React / Next or other JS frameworks. I prefer Vanilla JS which gives me direct control on my logic, without (overkill) abstractions. It’s OK for my small projects.

I tend to dedicate max 5 days of AI-assisted coding to a new project, spread over 2 weeks.

When I develop in VS Code with CoPilot, I deploy on Railway.app and my DBs are MySQL. When I develop on Replit, the default DB is Postgres, the project is then deployed on Replit.

Need assistance? Drop me a line.

I hope that this email gave you some inspiration for your own AI-assisted coding endeavours.

I’m at your disposal if you need help on a specific project.

I can help you craft your very own Personal Software.

I provide ad hoc and recurring consulting services. Simply reply to this email to start a conversation.