TLTD #20 - How AI Changed the Way I Work

The Leader's AI Stack

2024 was probably the year Generative AI changed from a party trick to a truly useful workflow, both at work and in my personal life. Like many, I’ve experimented with a bunch of AI tools since ChatGPT was released in 2022 but it just never really clicked for me until last year.

In this article I’ll share which tools I use the most and how I use them. Make sure to read until the end for an exciting announcement! The infographic below shows a high level summary. Feel free to download the full-resolution image here.

From searching to answering

The biggest challenge Perplexity has ahead of it is how to turn their company name into a verb.

People are used to “Googling” things but I don’t do that as much anymore and have taken most of my search needs to Perplexity, which its founder describes as an answer engine. It allows me to jump straight to reference-backed summaries instead of sifting through countless links to piece together the answer I’m looking for.

The key impact for me is how much faster I can conduct research and dive deeper into specific areas. Things like fact-checking, case study references and even product research can be done much more efficiently and reliably.

As a concrete example, I was working on a project where I needed to research AI Governance frameworks. In a short amount of time, I had a summary that reliably outlined the current state, developments per country/region and global trends. There’s no turning back now!

A Stack for Different Thinking Modes

During this full year experiment, I realised that different AI tools excel at different tasks and modes of thinking. I’m calling this the "Leader's AI Stack" — a way to leverage key AI products that augment different aspects of leadership work.

Strategic Thinking Partner: Claude

Claude was the first of the big LLM brands to add projects as an organisation mechanism. As such I started gravitating towards it more and more. Claude allows me to add documents as part of the project’s memory, grounding every conversation to that context. This is key in minimising hallucinations as well as providing helpful, relevant answers.

When I need to pressure-test ideas, explore strategic alternatives or create content, Claude has become my go-to AI buddy. Its ability to maintain context and engage in nuanced discussion makes it invaluable for:

• Brainstorming new initiatives

• Drafting strategic proposals

• Writing code

• Decision support

Knowledge Synthesis: NotebookLM

When I first caught wind of NotebookLM, it blew my mind! The ability to turn research papers and technical documentation — or anything, really! — into engaging podcasts has completely changed how I consume content. It's like having a personal research assistant who can:

• Convert complex documents into conversational format

• Generate insightful questions about the material

• Create summary podcasts for consuming on my daily commute

Multi-modal Communication: ChatGPT

I listen to audiobooks. A lot. I’m very good at consuming content that way so it’s common for me to do that during my commute or during walks. The trouble is that if I want to take notes, I need to open up a note taking app or a recorder app and save it that way.

With ChatGPT's advanced voice mode, I can create a new chat about the book I’m reading and simply tell it all the things I find interesting. As I’m reading a book, I’ll stop at sections where I’d normally place a bookmark and speak to ChatGPT instead.

This allows me to have a full discussion about the book at a later time, leveraging both what ChatGPT knows about the material as well as my own notes — all through voice while doing other tasks.

In addition, ChatGPT is my current choice for generating visual content for any documents I’m producing (including the articles in this newsletter).

When privacy matters

The tools I’ve described so far are great, but using them for work can be tricky. You have to be very careful not to input any confidential or private information which could be used to train these tools or that risk being leaked to the public.

Fortunately, there are free* alternatives such as LMStudio and Ollama. Both allow you to run Large Language Models on your laptop and each have their own trade-offs.

Since the models are running on your own machine, locally, the content remains as private as you want it to be. I use this whenever I need to analyse confidential information that isn’t trivial to anonymise.

* LMStudio is free for personal use but if you use it for work in a company beyond a certain size, you will need a commercial license. Ollama is entirely free.

Second order impact: clarity of thought

While there have certainly been productivity gains, the real value I’ve gotten from adopting these tools is to improve my own thinking.

Large Language Models are only as good as the questions and context you give them. Being able to reason through a complex scenario, describe the nuances of work relationships and express different view points in written form is a skill in and of itself. This is invaluable.

Ask yourself this: if AI tools went dark tomorrow, could you still function, however slower that might be? Through diligently working on my own reasoning and writing skills, I can confidently say yes.

Staying curious

The secret ingredient to making all of this work isn't the tools themselves, but rather being curious. I started my career writing code, building desktop computers and solving technical problems. That same curiosity that drove me to understand how systems work is what led me to explore and experiment with these AI tools until I found workflows that enabled me to be more effective.

Remember when Git first came into the scene? Or when cloud computing emerged? Each technological shift required us to be curious enough to explore, embrace failure and persist to figure out how it all fits together. AI tools are no different.

I encourage you to approach these tools with an open mind and curiosity-first approach. Start small:

• Pick one repetitive task in your workflow

• Experiment with different tools and approaches

• Share what you learn with your teams, friends and family

The stack I've shared isn't meant to be prescriptive nor exhaustive — it’s only a starting point. Your own AI stack might look very different based on your role, preferences, and challenges.

After sharing these insights with many of my contacts, one consistent piece of feedback was this: people want a structured way to develop these capabilities without the months of experimentation I went through…

Introducing “Mastering AI for Personal Leadership”

The trouble with Generative AI tools and how quickly they advance is that it can be pretty daunting to get started. You might spin the wheels for way too long trying to find something that works for you: which tools, how to craft prompts for certain tasks, which tasks are best suited for which model, how to fit them into a cohesive workflow…

This is why I’m launching a new cohort-based course: “Mastering AI for Personal Leadership”. I’ve done the hard[-er] yards so you don’t have to. This is a 3 week course that includes 6 x 1-hour live sessions packed with hands on advice to help you effectively integrate these tools into your work and/or personal life. The first cohort kicks off at the end of March.

I’m currently working on the course materials and would love to get your feedback to understand what would make the course most useful to you.

I'm offering early survey respondents exclusive discount codes. Plus, you'll help shape the course content and reserve your spot in the first cohort. You can access it here or through the button below

Looking Ahead

As 2025 marches on, I expect the integration of AI tools into leadership workflows to accelerate. But the fundamentals won't change: curiosity, experimentation, and continuous learning will remain our most valuable assets as technical leaders.

What's been your most surprising AI tool discovery? Share your experience in the comments below!