Last quarter, our CTO gently suggested, “Time for you to go to AI school”
Apparently, prompting ChatGPT with “Make me a business strategy” and then being disappointed by the vague response wasn’t exactly a shining example of AI leadership. Fair enough.
The takeaway? You don’t need to be a machine learning expert to work effectively with AI. What you need is curiosity, a tolerance for feeling inept, and a willingness to improve through iteration. Here’s what I’ve learned from my journey—less a masterclass and more a meandering learning curve—with one of the most powerful tools now at our disposal.
Lesson 1: Stop Being a Lazy AI User (Yes, I Was Guilty)
At first, I treated AI like a vending machine: insert a prompt, get a result. It didn’t work.
Instead of getting a useful draft of a simulation client pitch, I got generic fluff that read like someone googled “AI for finance” and stitched together a LinkedIn post. I realized that large language models (LLMs)—think of them as context-sensitive autocomplete engines trained on much of the internet—only work well when you guide them thoughtfully.
Let me be specific. Instead of typing:
“Write me a pitch for our simulation platform,”
I now start with:
“Write a 200-word pitch for a simulation platform designed to help mid-tier banks stress-test credit policies using synthetic data. Our typical user is a Head of Lending or CRO in Southeast Asia. The tone should be consultative, and it should emphasize strategic foresight.”
Then I iterate:
“Add a real-world example of how one bank used it to test risk-based pricing under a volatile interest rate scenario.”
“Make the second paragraph less technical—we’re presenting this to a room of non-quants.”
With just a few follow-ups, I move from unusable to presentation-ready. The key: treat it like a smart, fast, but occasionally tone-deaf research analyst who needs coaching.
Lesson 2: Redefine Productivity (It’s Not About Tasks Anymore)
We all love checking boxes. “Draft partner MoU? Done.” “Update pricing strategy deck? Done.”
But AI changes that dynamic. Now it can draft the MoU, synthesize feedback from four client calls, and suggest how to price a new module in our simulation suite—all before I’ve finished my coffee.
So what’s my job? It’s not checking boxes. It’s making decisions, adding insight, and connecting dots across the business.
Take this real example:
Instead of spending an afternoon drafting simulation learning objectives for a client in Vietnam, I asked AI to create a starting draft based on the client’s sector, our modules, and regional regulatory context. I then used my time to refine the nuance and tailor the delivery approach. My time shifted from execution to impact. In the new world of AI, your job is impact, not tasks.
Lesson 3: Understand Context Windows (Or: Why AI Suddenly “Forgets” Things)
AI’s “short-term memory” is limited. Technically, it operates within a “context window,” measured in tokens (roughly words or parts of words). Depending on the model, this can be anywhere from 4,000 to 128,000 tokens.
So when I’m developing a 40-slide client deck with multiple rounds of input, AI can start to lose the thread if I push too much at once. I’ve learned to structure the conversation like this:
First: “Create an outline for a training module on portfolio stress testing for SME lenders in the Philippines.”
Then: “Now write Slide 1 in bullet format, followed by detailed speaker notes.”
Then: “Now write Slide 2…”
You get the idea. It’s like working with someone sharp but easily overwhelmed—feed information in chunks, and clarify as you go.
Lesson 4: Follow the Tangents (Some Are Goldmines)
The magic often happens when I deviate from my agenda.
I once started a prompt:
“What are key drivers for simulating consumer credit portfolios during periods of interest rate volatility?”
Somehow, that led us down a path analyzing how banks are experimenting with price sensitivity and adverse selection.
The real insight? The best AI sessions aren’t transactional. They’re exploratory. Stay open to where the conversation leads.
Lesson 5: Embrace Variability (It’s a Feature, Not a Flaw)
Ask AI the same question twice, and you’ll often get two different answers. That used to annoy me—now I rely on it.
For example, when crafting the positioning for a new AI feature we’re adding to our simulation platform, I asked:
“Write a product description for bank executives explaining how our Agentic AI helps analyze simulation results.”
One version emphasized cost efficiency. Another framed it as a strategic advisor. A third focused on usability and transparency.
Each gave me something valuable. I merged the best bits into our actual pitch. It’s like having three consultants in the room—each with a different lens on the same problem.
The Bottom Line
Six months ago, I thought AI was a shiny toy for big tech. Now, it’s part of my daily workbench. I use it to think, draft, strategize, and sometimes just sanity-check my assumptions.
It hasn’t replaced judgment or creativity—but it’s amplified both.
And no, I haven’t become an AI expert. But I’ve learned to ask better questions, follow interesting threads, and get more done with less friction. And that’s leadership in the age of AI.
P.S. Yes, I used AI to draft parts of this post. How did I do it? I drafted the blog, then I asked two different LLMs to help me edit and refine. Then, I iterated a few times and here we are!
Interested in Learning How to Optimize Your Data with AI?
More Resources
– [Anthropic’s Guide to Effective AI Conversations](https://docs.anthropic.com/en/docs/build-with-claude/prompt-engineering/overview) – Technical but accessible guide to better AI interactions
– [OpenAI’s Best Practices for Prompting](https://platform.openai.com/docs/guides/prompt-engineering) – Practical tips for getting better AI responses
– [Coursera: AI for Everyone](https://www.coursera.org/learn/ai-for-everyone) – Non-technical introduction to AI concepts
– [MIT OpenCourseWare: Introduction to Machine Learning](https://ocw.mit.edu/courses/electrical-engineering-and-computer-science/6-0002-introduction-to-computational-thinking-and-data-science-fall-2016/) – Free, more technical course if you’re feeling ambitious
– [Banking Dive: AI in Financial Services](https://www.bankingdive.com/news/artificial-intelligence/) – Industry-specific AI news and insights