LLM Top 1% AI Strategies to Enhance Decision-Making in Your Business

In today’s Future Friday…

• 🪨 4 pillars to Forge Your Code into a Rock-Solid Business Asset

• 👣 9 Inevitable Steps to Propel Your Business into the Coveted 1%

• 🤖 ChatGPT’s New Custom Instructions Feature

• 📚 Unearth the Top Libraries and Prompts that Maximize Output with Code Interpreter

• 😉 And More…

If someone forwarded you this email, subscribe and join 5,000+ founders getting actionable golden nuggets tailored to make your business more profitable.

In our last Deep Dive, Benson shared unique gems about Code Interpreter that you can start implementing to boost your productivity to the next level.

Now the question is: how do I take Tmaximum advantage of this new feature?

After this Future Friday edition you’ll have the tools to really squeeze even the last drop of juice of Code Interpreter, and not only use it like your Data Scientist, but a top 1% one.

4 pillars to build a strong base for coding:

You know what they say: the deeper the base, the steeper the ascent.

That’s right, if you want to use Code Interpreter like the 1%, you must have your code straight.

This case is 100X more relevant if you’re coding with LLM-based coding tools such as ChatGPT, GitHub Copilot or Amazon CodeWhisperer, so make sure to apply these best practices to write incredibly better code.

4 pillars to build a strong base for coding:

1. Don’t trust if you can’t verify

2. Iterate one chunk at a time

3. Provide feedback

4. Clean the LLM context

⚒️ 9 Inevitable Steps Towards the Top 1%

If you are using ChatGPT as a Data Scientist with Code Interpreter, make sure to squeeze every last drop of its potential. 

You must be aware of them, otherwise you’ll end up getting mediocre results.

There are 9 main stages in this journey that you’ll need to consider:

1. Data Acquisition: Choose and utilize an appropriate library to scrape and collect the necessary data from the web or other sources.

2. Data Understanding: Inspect the structure, size, and general characteristics of your dataset to familiarize yourself with its content.

3. Data Cleaning: Handle missing values, duplicates, outliers, and other inconsistencies to ensure your data is clean and reliable.

4. Exploratory Data Analysis (EDA): Analyze your data to identify any trends, patterns, or relationships between different variables.

5. Data Preparation: Engineer new features from existing data to improve the performance of your machine learning models.

6. Modeling: Select a suitable machine learning library and model to train and cluster on your data, based on your specific problem and requirements.

7. Evaluation: Evaluate your model's performance using appropriate metrics and interpret the results to understand the model's strengths and weaknesses.

8. Visualization and Communication: Use suitable data visualization libraries to create plots and charts that effectively communicate your findings.

9. Iteration: Based on your evaluation, iterate on your model or data preparation steps to improve your results, and repeat the process until you achieve satisfactory outcomes.

🏆 Golden Nuggets

There are many different libraries you can use in each of them to take maximum advantage of the stages of scraping, exploration, visualization, and machine learning.

ChatGPT’s Code Interpreter is unaware of the libraries it can use, but you can manually upload missing libraries, and get it to extract and then use them.

Check out this tweet that shows what to look for in each stage and therefore, how to choose the libraries. To mention a few, you’ll be using libraries such as Selenium, Scrapy, Pandas, Scikit-learn, Tensorflow or Seaborn.

Now, to put the 🍒 on the top: prompts to enlighten your path to use ChatGPT as the Top 1%. Alright. Let’s get into them: