There’s an old saying about “Jack of all trades, master of none”, but typically we forget to add the last part “but oftentimes better than a master of one”. Generative AI is making a fundamental shift in the favor of generalists as we speak. 

In this article I’ll be talking about how I personally use AI rather than generalized viewpoints. I’m assuming the reader is familiar with ChatGPT, Bard, Microsoft Copilot or similar. If not, have a look at the links at the bottom (or do any random Google search. It’ll be hard to avoid hitting introductory articles about the tools 🙂).

In my role as a managing director, a significant part of my week involves face-to-face interactions. I engage in various types of customer meetings, which include sales discussions, negotiations, building customer relationships, and matters related to current projects. The meetings centered around ongoing projects primarily concentrate on delivery aspects, with an emphasis on delivering value to our clients through our active software development projects. These sessions often involve problem-solving, exploring alternative solutions, conducting cost-benefit analyses, and making decisions regarding operations.

My background is in both technical and business oriented leadership roles for the past 20+ years, with the last five years focusing on what you could broadly describe as intelligent solution development. They tend to incorporate data measurement and acquisition, data driven decision making, lean six sigma analysis, machine learning and utilization of artificial intelligence in the context of software development, and software platforms..

The generative AI tools I’m referring to can be found at the bottom of the article.

In my work, the most scarce resource is time and mental capacity. It manifests itself in various different ways, but fundamentally if I can do more (quantity or quality), or do the same in less time or less spent mental capacity it’s a definite win. 

I’m listing the top use cases where I’m using generative AI on an ongoing basis at a level that clearly gives value. These are no longer “trials or efforts done just for the fun of it” but actual work-critical activities:

Contract and legal text analysis:  

(note: not only syntax of text but the actual semantic meaning)

Legal text tends to have a good structure and clearly defined articulation per context. When making agreements in the software consultation industry you tend to always have the same key negotiation points (pricing terms, indemnification, limitation of liability, etc) you want to double check, and generative AI is very good at spotting the differences and highlighting them for you.

Notes and content creation:

Minutes, memos, diaries, drafts for emails and presentations. 

Driving and you’d want to articulate content? My personal feature of the ChatGPT iOS client is the ability to “hold down the microphone key” for as long as you need while you are dictating (or rambling) for ChatGPT. It is remarkably good at capturing your guidance and summarizing your thoughts per instructions. It does not hurt that it is absolutely tireless and chipper even when you ask it to summarize the same thing for the 11th time.

Mind map and flowcharts:

With AI diagram plugin you can turn your notes and text into mind maps. 

Often visualization is critical to be able to organize data and bring clarity to any topic. With the plugin capability you get a multidirectional multimodal ability (phew - what a sentence) to transition from verbal dictation to text to visual representation - and back.

Language and writing assistance:

Translations, spell checking, tone of voice analysis and adjustments, generating baseline suggestions. 

For these to be actually useful on a daily basis you want to define the context for the generative AI on what you are expecting before actually adjusting any text, e.g. generating schedules and instructions for padel tournaments vs. marketing material for your high tech company requires very different prompt engineering. My point is, it works - when you know how to define expectations as prompt instructions.

Presentation imagery:

Remember the awkward Shutterstock shuffle where you try to find pictures that are not overly cheesy and still represent the idea you want to convey? With a combination of creative concepting tools, e.g. DALL-E-3, and modification/iteration/enhancement based tools, e.g. Stable Diffusion, you can create marketing material pictures of equal quality that describe exactly what you want. Around 50% of our presentation pictures are generated and this number is increasing rapidly. I’m moving more towards a direction where I’m using a seed image as a starting point and then iterating with generative AI from there.

A better search engine than google when you don’t have the right keywords:

Often I’m seeking information from a particular person or book but don’t remember the name of either the book or writer 🙂 Describing the content of the presentation, plot of book, or key concept allows ChatGPT to suggest the most likely source for the information with uncanny precision. A showing example was when I was searching for a book by William Gibson on the topic of “AI in the form of nanotechnology, as a book that helps the main protagonist on her growth journey”. The suggestion from ChatGPT pointed out that I was probably confusing William Gibson with Neil Stephenson - a common mistake it emphasized with me - and the book I was referring to was Diamond Sky.  

These are my top 2023 use cases for daily use of Generative AI. I’m honestly excited to see what will change when I post my 2024 version. I wish you exciting times with generative AI and a relaxing Christmas break! 

Note on ChatGPT. I’m using the commercial version (ChatGPT4) and I have enabled beta settings. I typically have all of the following plugins enabled: AI Diagrams, wikipedia, Wolfram and Scholar AI.

I’m using extensive pre-instructions for ChatGPT to define work context, expectations of brevity, and style of communication as well as instructions on how to handle different language input.

Links:

www.midjourney.com
www.Openai.com
www.wolframalpha.com/
https://openai.com/dall-e-3
https://stability.ai/stable-diffusion