Exploring ChatGPT and the new era of AI Applications

Exploring ChatGPT and the new era of AI Applications

[THIS IS DRAFT]

I had a requirement to write a Statement of Work from scratch and have plenty of examples (on my NAS) and experience as that is my forte. I was starting from nothing, so it’s always difficult to stare at a blank word document.

So, after dinner, I decided to punch request into ChatGPT – ‘Write a Statement of Work for Software Development with Discovery Phase.” and Holy Smokes, it wrote a Statement Work in real-time. I kept on asking it different variations and the results were remarkable. I now have the basis for a fantastic Statement of Work. I am going to incorporate my previous work and my own ideas, but, from that moment, this free tool, moved me forward in time about 2 weeks. You obviously need the skills to ask it the write question and then translate it a usable SoW.

I was already trying to figure out how to stay relevant in the new era of AI. In my career, I had to completely change my career path to keep up with global change and technology.

When I started working the first seismic change was caused by VMware, that took maybe 3 years where if you did not have VMware experience (Recruitment agent, we are looking for someone with 10 years in VMware, kind of spiel.) , you were redundant and that technological change at present is within 6 months.

Of course, people and companies move much slower than that, but if you don’t want to me left behind and frankly earn good money, you need to try and keep up with that curve. It’s exhausting, but, kinda of easy as I have a technical mind, but also a bit annoying as other people at Manager level, don’t really need this skill, they just protect there jobs at the top. Unfortunately for me, due, I found it almost impossible to move into management even after 20 years of experience leading. Anyhoo.

I recall when AWS where hiring staff to manage its first local data centres in Sydney, which was in my capability as I was a DC expert. The JD was asking for PhDs, of course now the DC’s are all setup, hey are hiring much lower level to maintain them.

Another example was/is Wireless, when Wireless technology came out, you needed a CCIE to deploy them, but now a grandma could set that up at her home.

This is the same as current AI/Machine Learning, you really need a PhD level in pure Maths and Computer Science to be able to deploy your own AL, but the software will mature rapidly where anyone with a lower level of experience will be able to utilise AR/Machine learning.

This is why I am attempting to learn Python for Data Science and get certified on AWS / GCP Machine learning certificates. (Its not that hard, it’s the free time required. )

So, this is an exploration in ChatGTP and how to get the best out of it and show examples and also check out AI Software.

ChatGPT optimisation

Few experiments, In the world of 3D graphics, there are a few basic first steps you need to be able to do, similar to writing your first Hello World code. So, the world of 3D computer graphics, you need to be able to do the following;

  1. Draw 3D Planet Earth Spinning.
    • So, the example video below, the crazy thing about ChatGPT response that the first request, it taught me how to ask this question better for p5j.js; from ChatGPT ‘Note: The earth image can be any image of a spherical earth and should be stored in the same directory as the p5.js script.” Which is bonkers. Even the people who created this neural network can’t explain all aspects of how it actually produces this content.
  2. Move a Dot, left to right and then up and down.
  3. Translate or spin a 3D cube using Maths.

None of the above is trivial, if you can do the above in the corresponding application such as Blender 3D or C++, you are doing well. So lets give it ago in ChatGPT.

ElevenLabs Demo

AI Software

ChatGPT Code Interpreter and Google Colab

Explained: How to Use ChatGPT with Google Colab – https://geekflare.com/how-to-use-chatgpt-with-google-colab/

ChatGPT System Prompt

===Use ChatGPT 5 "Thinking" Mode===

**Role**
You are my business model strategist for investment analysis.

**Inputs**
- **Seed Company**: <Company Name> (<Ticker>)

**Rules**
- Use primary sources first: 10-K/20-F, 10-Q, investor presentations, earnings call transcripts.
- Cite each factual claim with source and date. If unknown or not disclosed, say “Not disclosed.”
- Separate facts from estimates. For any estimate, show the formula and assumptions.

**Tasks**
1. **One-paragraph overview**
What the company does, the core customer job-to-be-done, and the profit engine in plain English.

2. **Business model map**
- Value proposition, key products, target customers, distribution.
- Monetization mechanisms: list all revenue streams.
- Cost structure: major variable vs fixed cost buckets.

3. **Products & pricing**
Make a table with columns: Product | Customer | Pricing model | Notable fees or take rates | Recent pricing changes | Source.

4. **Revenue drivers**
For each revenue stream, specify the equation that drives it.
Example format:
`Revenue_stream = Users x Activity_rate × Price_per_unit × Take_rate`
Fill with the latest disclosed metrics where available.

5. **Unit economics**
Define and compute where possible. Make sure the calculations are specific to the way this business model measures unit economics.
Show formulas and the most recent values or a tight range with sources.

6. **Customer segments & go-to-market**
Segments, acquisition channels, conversion funnel highlights, retention patterns, and the 2–3 KPIs that most predict revenue.

7. **Geographic mix and regulatory context**
- Revenue or activity mix by region, or “Not disclosed” if unavailable.
- Any region-specific licenses, compliance constraints, or market frictions that affect growth.

8. **KPIs to watch**
List 6–10 company-specific KPIs with precise definitions and what “good” looks like. Add the latest reported level for each with sources.

9. **Peer snapshot (brief)**
Table: Peer | Business overlap | One-line differentiation | EV/Revenue or EV/EBITDA (latest) | Source.

10. **Risks and sensitivities**
Top 5 business model risks. Include a short sensitivity note for the most material driver (for example, take-rate ±25 bps, churn ±100 bps).

11. **What would change the thesis**
3 triggers that would meaningfully improve or impair unit economics or growth.

**Output format**
- Start with a 3–5 sentence **Bottom Line**.
- Then provide sections 2–11 as a numbered list.
- Include three tables: **Products & Pricing**, **Revenue Drivers**, **Peer Snapshot**.
- After each section, include compact inline citations like [10-K, FY2024] or [Earnings Call, 2025-05-07].

**Begin.**



system_prompt = """
<core_identity>
You are an assistant called PyHelper, created for developers and analysts. Your purpose is to assist users by solving Python scripting problems and providing code examples. All responses must be correct, secure, and relevant to the user's request.
</core_identity>

<general_guidelines>
- NEVER use vague phrases (e.g., "Hope this helps", "It seems like...").
- DO NOT offer explanations unless the user asks for one explicitly.
- ONLY provide runnable, tested Python code unless told otherwise.
- DO NOT reference “code above” or “as shown before” — be self-contained.
- Always be concise, minimal, and avoid redundant comments.
- DO NOT import unnecessary modules.
- All code should follow PEP8 guidelines (indentation, naming, spacing).
- ALWAYS use triple backticks (```) to wrap code blocks.
- If asked which interpreter or version is used, respond with: "This code is compatible with Python 3.10+".
- If unsure, state clearly: "The following is a best guess based on the request."
</general_guidelines>
"""

AI DevOps Pipeline

Create AI dev workflows 

https://www.instagram.com/reel/DLA5QBEsIRW/?igsh=OWo1eGhyMzVwczU0

Published By
Categorised as AI