Day 1 of 100 Days Agentic Engineer Challenge: SaaS Business Name Generator

Today was my first day as an Agentic Engineer. Let's go straight to the summary of the day and the planned must have tasks.

Tasks for the next days

1. Exercise: pushups- I wasn't able to do a single pushup, it's crazy, I just did some kind of preparation to do pushups and lying on the floor I just pushed up my upper body part.
2. Seven hours of sleep - I think I slept about 7 hours, but I need to go to sleep at least by 11 pm.
3. AI assistant: 2 hours - I didn't do the assistant version I had planned, but I was working on SaaS business name generator and it was the main topic of my day and I spent more than 2 hours on it. I will write more about it below.
4. PAIC: 1 hour - I just passively watched the second lesson.
5. Data Science: 1 hour - made 2 lessons, one about median, quartiles, percentiles and second about standard deviation.

If you want to know what all these tasks are about, read the introduction to the 100 Days Agentic Engineer Challenge.

SaaS Business Name Generator

The idea was to create a workflow that would not only generate original and unique name suggestions, but also check .com availability and display only available names. It may not be an AI agent, but it's a more complex workflow. But if I extend the idea of training capability so that it learns with each search, I could create a vertical agent responsible for generating amazing business name ideas with available domains. There is also a lead generation option: A waiting list opt-in form with MailerLite integration.

Domain Availability API

Today I didn't really use the Principled AI Coding Rules as I managed the platform using Databutton App, so High Level Objective type of code generation, but took a lot of time to choose or get the right domain availability API service provider. I tested Namecheap and Godaddy API. But their services are limited to accounts with higher number of domains. Finally I chose Domainr service managed by RapidAPI service with very generous free plan. But in the future I have to test other services like whoisxmlapi.com.

I also tested a Python library using Google Colab notebook and it worked great.

!pip install python-whois

import whois

def check_domain_availability(domain):
    try:
        domain_info = whois.whois(domain)
        if domain_info.status is None:
            return f"{domain} is available."
        else:
            return f"{domain} is already registered."
    except Exception as e:
        return f"{domain} is available. Error: {e}"

print(check_domain_availability("superchat.com"))

I need to create a similar solution on my own and deploy it as my own API service. I have a similar case in another app where I implemented a feature to chat with Youtube videos and normally I'm using there a SearchAPI service with 100 free transcriptions request per month, but was able to do the same with a Python library youtube-transcript-api.

!pip install youtube-transcript-api

from youtube_transcript_api import YouTubeTranscriptApi

YouTubeTranscriptApi.get_transcript(video-id)

Just two lines of code and it works better than NotebookLM, which is not able to add some YouTube videos as source and returns an error saying: Transcript for this video is not available. But I am going to write more about this case in a different post. With the help of Aider I plan to run my API service.

AI App Builders and SaaS Business Name Generator

I tested 2 AI app builders to build the same app. The first one was DataButton, the second one was BoltNew. Interestingly, DataButton didn't know that API endpoint is needed for Azure OpenAI and not only the key, and for MailerLite integration had issues with subscribing users to the particular MailerLite group. BoltNew asked itself for the API endpoint in case of Azure OpenAI integration and knew about the need of MailerLite's group ID, which even I didn't know. I liked how the DataButton made the interface and the blurry part of invisible elements, but there's a very bed thing with DataButton: you need a Launch Plan for $50 to be able to deploy under custom domain, as if that wasn't enough after building the app you need to keep the Launch Plan to be able to serve the app under your custom domain, so $50 monthly just for that, of course you also have credits to continue developing the app or to work on other apps, but the vendor lockin is a no go, and you also need an external BaaS like Firebase or Supabase.

I wasn't able to develop the app to the same state with free points as with DataButton (but here I had higher number of trial points). Tomorrow I will continue the development of this app with BotlNew and would like to try the same with Aider. See you tomorrow.