Issue No. 16
As I'm building this newsletter (and a podcast and YouTube channel) in the open, you will get updates on this project here from time to time.
I'm back from a hiatus of almost a year! Unfortunately, this project had to take a back seat due to a crazy period of work commitments and challenges on the home front. On a personal level, it has been a period full of challenges that have given me many occasions to reflect and, hopefully, learn.
I'm sure that, like me, your life is super busy, so I want to keep my posts short with no fluff, with the occasional deep dive on a topic.
I've been reflecting back over the last year, one of the things that has happened is the explosion of "AI," or rather, Large Language Models and generative AI.
Like with any technology, it is already being abused. From crappy SEO-optimized product blog posts (they used to be bad enough when marketers wrote them) to whole books being generated and swamping digital bookstore shelves.
I worry that soon we are going to be wading into an even worse Internet than the social media-hyped version we are already in.
On the other hand, I'm a geek, not a Luddite, so I have been playing around a lot with this technology, and I believe it should be part of every knowledge worker's workflow.
Some of the applications of this technology will give those who use it a distinct and unfair advantage over those who do not use it (or who do not master using it).
One way to use generative AI is as a thinking partner to ask us questions and test our ideas before turning our answers into varying lengths of content, which we can then apply to our final edits.
In order to bring this to life, I think a short example might be helpful.
An Example of Using AI to Help You Think
One of the things I try to do, and which I coach my teams to do, is to focus on outcomes rather than outputs.
I also use a specific definition of an outcome - “a change in human behavior that impacts business results.” This comes from the book “Outcomes Over Output: Why customer behavior is the key metric for business success” by Josh Seiden.
What we try to do is to have every initiative, or "project," named using the language of an outcome - the actual human behavior we are looking for. Additionally, every increment or slice of those "projects" should also use the same outcome-oriented language for the name of the increment.
There is SO much clarity when you can give your work titles like this. First of all, we benefit from having clearly articulated the "Why" of the work we are doing. Secondly, even the least technical person can quickly understand the work we are doing and what the team is trying to achieve.
However, coming up with short, impactful outcomes for our work turns out to be pretty hard to do in practice! Especially for a very technical project or spike where it is much harder to identify the user and link their behaviour to the success of the business.
As they say:
There are only two hard things in Computer Science: cache invalidation and naming things.
– Phil Karlton
In the case of coming up with short outcome statements, I have found generative AI to be super useful to ask me questions and summarise the results.
Here is an example of a prompt I'm still refining but I've started to use to be able to name and describe the work we do when I'm a bit stuck:
I'd like your help in updating some project descriptions to be written as an outcome.
The definition of an outcome is a change in human behaviour that impacts business results.
Each project description should clearly identify:
- the person or human or persona
- the behaviour change for that person
- the link to business results between their behaviour and the impact on the business
Please ask questions if needed and provide prompts to guide me so you can suggest an updated and short project description.
Do you understand?
Note: If you are using Chat GPT, there are radically different results between GPT 3.5 and 4. I have also found that there is a bit of variability in the results.
This helped me take a fairly technical feature
"biometric user validation"
In a minute with a few prompts, I had:
"Improved user experience and adoption through biometric user validation for passkeys"
Which I then edited to:
"Improve UX and adoption of passkeys by adding biometric user validation"
I know I could improve this with more time, but I hope you see how AI can help us when we're stuck and give us something good to use fast. For busy knowledge workers, this is really helpful.
And this is only one of the many uses of generative AI!
Best Practices to Use AI to Help You Think
Generative AI can be a powerful tool when used correctly. Here are some best practices when using it as a thinking partner:
- Clearly define the problem or question you want to explore before using generative AI. The more specific and focused your question, the better results you will get.
- Garbage in, garbage out - if your thoughts are flawed or incomplete, you will get subpar results.
- Don't rely solely on generative AI for all of your thinking. It should be just one part of your toolkit.
- Experiment. Use multiple iterations of the same prompt to generate different perspectives and ideas.
- Don't sound like a lame bot. Edit and refine the output before sharing it with others or using it in a final product.
I believe generative AI can be an effective thinking partner and we are already starting to see it used for even more interesting use cases like my friend Tim's use of AI in their product, Stellafai, which helps teams manage goals and OKRs.
Don't worry, this newsletter hasn't pivoted to being another part of the AI hype machine! I've got some good content planned around execution and leadership coming up. I just couldn't ignore one of the biggest developments in tooling for knowledge workers we have ever had!