In this week’s newsletter, I wrote an essay about bottlenecks. I also recap what I wrote this week about Microsoft’s shocking buyout offer, AI food that makes you feel sick, rapid data storytelling, Muse agent in Singapore, shopping for apples, and much more.
Weekly Recap
🔵 The most earthshaking news of the week for me was Microsoft asking 27% of its employees to take a voluntary buyout. If you are offered the buyout, you should take the money and run.
🔵 Avoid eating from food outlets that misrepresent their dishes with AI imagery
🔵 How fast can I do data storytelling with Excel and Claude? The data is a little grim though – Q3 retrenchments in Singapore have gone up.
🔵 You cannot design an AI agent or AI solution and assume it is sentient like The Vision.
🔵 Singaporeans can now use Meta’s Muse agent. Be careful though!
🔵 This great book by Ryan Holiday is still heavily discounted at the Kindle Store for USD $2.99. Just buy it!!
🔵 Shopping for apples turned into a thinking exercise about one’s value at work
🔵 The three most common reactions to the impact of AI on one’s work and career.
🔵 But don’t ask me for career advice.
🔵 Jotting down random ideas for my 2027 AI workshops. I also like to print out the quotes that I want to remember.
🔵 Apple iPhone Duos may include a free lightsaber.
🔵 Congrats to my daughter!
🔵 Another Electric Minds morning meetup. This is where AI enthusiasts in SG have the most interesting conversations.
🔵 I was in a good mood and decided not to ban this LinkedIn user.
🔵 Hit me baby one more time.
Sunday essay
This week, I stopped in my tracks when I heard what Boris Cerny (head of Claude Code) said in this YouTube video:
“And the thing that we did to get to the 8x (increased coding output from Claude engineers) is we systematically improve a bottleneck at a time. First, coding is the bottleneck and we throw Claude at it and Claude does the coding. The next bottle neck is code review and we have Claude do that. The next bottleneck is generating materials for GTM (go to market)… and so just like one step at a time, we unbottleneck every part of the process.”
I was gobsmacked! That was a great way to describe the problem-solving process.
Before this whole AI disruption, we had to spend days, months, and years “unbottlenecking” the major bottleneck in our workflows. Sometimes, the bottleneck just stubbornly stays there and everyone gives up. The bottleneck could be a thing, a process, or a person. Maybe you had a technical document that was hard to read. A process that had become fossilised. Or a person who was a fossil and the boss refused to fire this deadwood.
Off-tangent: When I was managing 120 marketers at Razer, someone asked me, “Ian, how do you motivate an unmotivated employee?” I said, “It’s hard to get people to change. It’s easier to just hire a motivated person.”
It sounds cold-hearted, but I learned the hard way that serious human bottlenecks cannot always be solved by counselling. No matter how many months or years I tried coaching and counselling such unmotivated people, they were always held back by.. their lack of motivation to change. They either embraced the status quo, saw their job as transactional, had personal issues that affected their colleagues, or they had little to no ambition to pursue their true desires. I inherited these people and it was painful to work with them.
So, now we are at The Great AI Reckoning.
AI is like a bulldozer in the problem-solving process that can dislodge deeply embedded bottlenecks and fling them away. There are a few major types of bottlenecks that I can identify and it’s clear how AI is impacting them:
🔵 Old technology – paper forms, physical cabinets, printed dollar bills. These were replaced by digital forms, cloud storage and QR payment codes. Apps also age. Microsoft PowerPoint has hardly improved its feature set since 2007 when I joined Microsoft. That is why young people prefer to use Canva. And now we can create or reverse-engineer any type of software with AI…
🔵 Lack of tool knowledge (tool skills) – Most people never learn to use more than 10% of the features of Microsoft office apps such as Word or Excel (or new apps like Canva). There is little investment in tool training in many companies. That is why many processes are stuck in the past – the user simply does not know how to apply the technology to the problem. They also don’t know how to ask Gen AI to teach them how to use their existing tools.
🔵 Inability to identify valuable actions and outcomes. This is why there are billions of unnecessary daily meetings, marketing campaigns that fail on arrival, events which nobody cares about, and output that people don’t want to pay for. Many people assume that just because they have put much effort into creating something, it is worth something to another person. I also apply this to my AI learning – there are a million things to learn, but I spend time learning AI techniques that would be useful to the layman so that I can teach them the same.
🔵 Wrong person for the job. This is self-explanatory. The poor fit could be due to aptitude, skills, attitude, or experience. Unfortunately, many people are in the wrong job, and they may not even know it. They’d much rather ask AI where to go for lunch than to ask AI what are the flaws in their daily work.
Solving problems is core to every job worth doing. For centuries, we workers have focused on solving the problem that the job wants us to do (for example, a cashier’s job is to handle cash transactions). Along the way, we realise that the bottlenecks in processes, people, and technologies needed to be improved, so we optimised them the best way we could.
What AI has done is to vastly augment and accelerate the fixing or removal of these bottlenecks – that is why work has become more productive while job losses are mounting. If I notice a bottleneck in my work today, I will spend time to remove or mitigate it with either human resources or create an AI solution on the spot.
In any age, we must ask ourselves:
Are we creating bottlenecks, or are we ourselves a bottleneck?
Are we able to fix the bottlenecks or do we let them slow us down?