On Feb. 19, 2026, the ASSP Artificial Intelligence (AI) Task Force released its white paper, “AI and the Evolving Role of EHS Professionals.” To expand on the paper’s key themes, task force members are sharing their unique perspectives in this new weekly series in ASSP News.

Bootstrapping AI for Data Analysis, Microlearning and More – My First Attempts
By Saif Ahmad, senior EHS specialist, Kuwait Oil Company

I’ll be very honest from the start: I’m quite new to ASSP. I joined only last November, and before I could even learn everyone’s names, I got an email saying ASSP was forming a task force on the use of AI in environmental health and safety (EHS). As someone who’s always been a curious safety pro, an AI enthusiast and someone who’s been waiting for an excuse to visit the United States, I thought, why not?

And that’s where my “bootstrapping AI for EHS” journey began. Not with a budget. Not with formal training. Just interest, stubbornness and a middle-class Indian household instinct for thrift. I didn’t want to spend a single penny. 

My first attempt: Building an ML model with zero coding skills
I wanted to build a machine learning (ML) model to extract insights from our corporate EHS data. The challenge?

• No coding background
• No deep statistical skills
• And being, well … cheap

Here’s how the adventure unfolded:

1. Gemini: The beginning
I started with the Google Gemini CLI. I downloaded it, explained my problem and asked it to write a program using basic ML algorithms. It gave me something workable, but I knew Gemini isn’t famous for DevOps-level coding skills (sorry Google).

2. Claude: The reviewer
Claude writes cleaner, more structured code. But the free-plan context window taps out fast. So instead of starting from scratch, I uploaded Gemini’s code and gave Claude my requirements in one big shot.

Claude responded like a seasoned reviewer, pointing out flaws and suggesting improvements. After a few rounds (until the free token meter started blinking), we ended up with a refined version without sharing confidential company data on any AI platforms.

3. ChatGPT: The summarizer
Then I copied the whole Gemini–Claude exchange into ChatGPT and said, “Summarize this so I can restart the conversation on Claude without triggering the message cap.” ChatGPT became my context preserver, letting me hop across tools without losing momentum.

4. Tool-hopping cycle
Before I knew it, I had a tested freeloader AI workflow: Draft with Gemini, refine with Claude, summarize with ChatGPT. Repeat.

Did I fully understand what I built? Eh. Did it generate useful insights? Surprisingly, yes. It identified underlying accident trends, such as months when incident likelihood increases. Once these patterns surfaced, senior safety managers could point out the root causes behind them.

That’s when I realized: Free AI tools are serious enablers for seasoned professionals.

My second attempt: AI-powered microlearning for safety
In EHS, most of us grew up with the legendary “two-hour awareness session.” The instructor talks, we listen (or at least pretend to), and the coffee kicks in harder than the content.

Important topics like Working at Height or Confined Space Entry deserve more than endurance trials. So I thought: Why not try microlearning with a little help from AI?

Here’s what I did:

1. Took a safety topic and stripped it down to essentials using few-shot prompting in ChatGPT
2. Turned it into multiple 10-minute podcast-style conversations using NotebookLM
3. Used the multi-language option so our diverse workforce could listen in their own languages
4. Added a short explainer video (still English-only … for now)
5. Asked Claude to create a quiz “artifact” that gets harder as participants succeed
6. Captured quiz results to gauge effectiveness at both the individual and organizational levels
   

And the best part? Still a freeloader.

This doesn’t replace traditional training. It complements it by giving people more personalized, engaging ways to learn, reflect and strengthen safety awareness.

What I have learned (so far)
Working across the ASSP task force and in my own experiments, a few things have become clear:

• AI isn’t an elixir or a magic wand. It fits into your work system, not the other way around.
• Mindset is the biggest barrier. People expect too much, fear too much and act too little.
• We’re at the “peak of inflated expectations” in the Gartner Hype Cycle. AI won’t solve everything, but it will help us think better.
• AI won’t take your job; someone using AI will. At least for now.
• Context engineering is the new prompt engineering. Just when I started to understand prompts, the world upgraded.
  

And most importantly:

• Free tools + curiosity = real EHS value. You can start small, experiment safely and still create something meaningful.

Why this matters for EHS
Our industry deals with inherently high-risk environments. Yet thanks to strong controls and mature systems, incidents have plateaued. AI helps us identify patterns faster, communicate smarter and investigate with more clarity.

AI won’t replace the safety professional. But a safety professional who uses AI becomes an even greater asset — and soon, that may be the expectation.

That’s why I’m excited that ASSP took a bold step into this space. Safety is always shaped by new technologies that modify behaviors and create new challenges. In many ways, AI behaves like a new species. It hallucinates, negotiates, convinces and occasionally manipulates. This new species brings new risks, and who better to manage those risks than safety professionals?

Final thoughts
I’ve built imperfect models, hacked together learning tools, and bounced between free platforms more times than I can count. But through all the slipping and tripping in my AI journey, I know this much:

• You don’t need to be a coder.
• You don’t need a budget.
• You do need curiosity, critical thinking and the willingness to experiment.
  

If you’re an EHS professional curious about AI, start small. Summarize a document. Build a quiz. Analyze a spreadsheet. Create a micro-module. And if your first attempts look like duct-tape engineering, welcome to the club.