How to apply AI and automation in your business
“You never change things by fighting the existing reality. To change something, build a new model that makes the existing model obsolete.” Buckminster Fuller
Previously at the Volcano Base I'd been reviewing the first 3 months of being in business. Since then, I've been seeing patterns in how companies go about applying AI and automation.
Mission Briefing
Two approaches to applying AI and automation
Everyone’s talking about AI, but when it comes to actually using it in your company, things get a bit… squishy. Should you start with a grand vision? Or just pick off the annoying stuff first? The good news: there are two solid paths. The bad news: you actually have to pick one.
The Top-Down Approach: Strategy Nerds Assemble
This one’s for the planners. It starts with leadership setting out the company’s goals. Think “reduce customer churn” or “scale without hiring more humans", and then figuring out how AI or automation could help. Maybe that’s better forecasting, maybe it’s smarter data analysis, maybe it’s a CRM that doesn’t require a PhD to use.
It’s deliberate, strategic, and can be slow. You’ll probably need consultants, workshops, and a bit of soul-searching. But if you want alignment, scale, and long-term value, this is the way.
The Bottom-Up Approach: Quick Wins for the Win
This one's scrappier. It starts with your team - the people who actually do the work - flagging stuff that’s repetitive, boring, and obviously ripe for automation. Then you weigh up: how much pain does this process cause? How easy would it be to automate?
Best practice: start with something that’s high value and low effort (yes, they exist). You’ll get quick wins, build confidence, and maybe even free up some time to think more strategically later.
Not sure where to start? Ask your team: “What’s the one thing you do every week that makes you want to scream into a cushion?” Then go from there.
Classified Intel
Some interesting stuff I discovered on my adventures...
OpenAI Academy - Learn the building blocks of AI
OpenAI’s new learning platform offers structured, interactive lessons on how to build with AI - great for anyone trying to get from “what is a token?” to actually deploying useful tools.
Why it’s useful: Whether you’re taking a top-down strategy or bottom-up experiments, you’ll need people who understand how AI works. This is a good place to start levelling up your team.
LM Arena - A sandbox for testing and comparing LLMs
LM Arena is like a fight club for large language models - compare responses from different LLMs side by side and see how they perform in various tasks.
Why it’s useful: Before locking in a model or vendor, it’s worth testing what works best for your use case. This tool helps demystify model behaviour without needing a PhD.
Building Effective Agents - Lessons from Anthropic’s engineering team
https://www.anthropic.com/engineering/building-effective-agents
Anthropic shares the practical stuff they’ve learned while designing AI agents - from managing state to breaking tasks into chunks that don’t melt the model.
Why it’s useful: If you're building custom automations or agents, this gives you a peek behind the curtain of what actually works in production.
Your data is not shared. Unsubscribe with 1 click.
Member discussion