Practicum Challenge: Blueprints for Enterprise AI

[onlyphantom]

🌹 You can attempt either

1. Challenge (TECHNICAL, coding required) or
2. Challenge (CONCEPTUAL, non-coding)

1. TECHNICAL Challenge: Build a natural language stock screener!

Make a copy of this code template I have created for you on Google Colab.

Use code snippets provided in the course material:

• Agentic Patterns course on Sectors Bulletin
• Any code you find in this Repository that might be good reference
• Re-watch the MP4 recordings of the Blueprints for Enterprise AI workshop

Instructions

The Code Template on Google Colab demonstrates a single-agent workflow. You have to modify the code in main() to demonstrate the use of a multi-agent AI workflow, featuring more than just natural_language_screener.

You may use any of the multi-agent patterns discussed in the workshop:

• The Manager Pattern
• The Skills and Tools Pattern
• The Sequential Chain Pattern
• Judge and Critic Pattern
• Parallelization Pattern
• Guardrails Pattern

Pick one of the above and modify the code. Run it from top to bottom and make sure the output of your multi-agent workflows is reasonable and reflects a real-life business case. Submit the link to your notebook.

If you're residing in Singapore / Malaysia, you are allowed to adapt the question to use financial stock data from Bursa Malaysia / Singapore Exchange pulled from Sectors Financial API (read the documentation).

Submission Format

To submit your completed exercise, open a new Discussion under 'Practicum: Blueprints', and paste a link to the completed work. This should be a Google Colab notebook. The output of each cell must be shown, whether they are intermediary steps or not. Make sure to 'Run All' from top to bottom before saving your work.

Specify your name and first 4 letters of your email at the top of your notebook for me to identify you.
Example: Samuel Chan | samu (if my email is indeed samuel@sectors.app)

🔴 IMPORTANT: DO NOT SHARE YOUR SECRETS. USE from google.colab import userdata to read your environment variables. Under no circumstance should you secret keys be ever visible. An inclusion of your key in the notebook will automatically disqualify your work, and for your own benefit, I will promptly remove your thread here on GitHub so that your keys are not stolen.

Grading

You will be given a Passing grade if your notebook meets all criteria:

1. Google Colab notebook with all cells executed from the top to bottom so the output of each cell is shown (for logistical and security reasons, I won't be running .ipynb notebooks on my local computer. Please submit a valid Google Colab notebook).
2. The use of a multi-agent AI system (more than one Agent working with each other to deliver a reasonable business use-case), including the natural_language_screener created for you. Your system must feature the use of the natural language endpoint (https://api.sectors.app/v2/companies/?q={query}) so your API Key usage shows up on Sectors internal log and is traceable.

You can discuss, and ask for help in Discord, but please do not share your answers or plagiarize. You are only cheating yourself of an education.

---

2. CONCEPTUAL Challenge

Not all participants of this workshop are of a technical background, and are able to write Python code at this level yet. To be more accommodating of beginners and non-technical participants, I have prepared an alternative challenge for you.

I will still need you to demonstrate that you have learned the concepts as intended by the workshop, so I will ask you to write either a short paragraph of your key takeaways or provide 3 - 4 bullet points summarizing your learning or insights from the workshop.

For example, you may focus on one or two of the following Agentic Patterns discussed in this workshop, and explain how a natural language stock research (+ stock screening) AI Assistant work in theory / in principle / in practice using either English or Bahasa Indonesia.

Submission Format

• Write up 3 - 4 bullet points on one of your own take and share it with your audience.
• You can include your thoughts on any key points, or any insight that you find particularly interesting when it comes to agentic ai tools capable of performing tasks by being instructed in natural language (English / Bahasa Indonesia)

Tag Supertype or myself (if on LinkedIn) or @sectorsapp (if on Instagram) in your post so that I am alerted to it and can review your submission.

[XterminatoR30]

Kevin Cliff Gunawan | work
https://colab.research.google.com/drive/1DdaRW7gXn3IIotQKPj7OIYg-fSYgskks#scrollTo=9-kWtr7tf3D0