I can still vibe code even when I'm exhausted

A observation on the future of productivity

When I was starting my first company, I had the very good fortune to share an office building with Sean Blanchfield, a seasoned Entrepreneur who gave me some great advice.

He told me that I shouldn’t just plan my day in terms of the amount of hours I had to work, but in terms of the energy I had to spend. From his 2011 blog post on the topic:

I imagine that there is a roughly fixed amount of energy that I have in any given day to spend how I will. However, different kinds of tasks, interruptions, conversations and people drain this energy at different rates

He then gave a few examples of the kinds of activities that can burn down energy at different rates, (measured what he called ‘Energon Cubes’ for fun):

• When I’m in flow, I use up one Energon Cube every 20 minutes or so.

• Phone call: 0.5 - 1.0 energon cubes

• Meeting: 2 Energon Cubes per 20 minutes

• Other people’s conversations: 1.5 Energon Cubes per 20 minutes.

• Heated emotions: 5 - 10 energon cubes per incident.

I really liked this insight, and ever since then I’ve made efforts to structure my days and weeks around my energy reserves.

I do reading in the morning, while I wait for my caffeine to kick in. I mostly do creative work before lunch and I mostly do meetings in the afternoon, when my creative juices are spent. I save all my boring, mindless admin works for Friday afternoons, for when my heart and my mind have left the office long before my body does.

One very interesting thing I’ve observed about my work patterns recently is that AI has helped shift some creative work from high energy consumption to low. It has enabled me to continue do creative work even when my personal batteries are low.

The first way this manifested was in the evenings. When 6pm rolled around, I had finished my meetings but I still had some emails to write, or a few slides to finish off presentation, I would trudge through it in a slow, painful, procrastination riddled extra hour of work.

AI has changed this. Describing the task to an LLM is easy. This can lift me past the procrastination hurdle and do the parts of work that I lack the energy for. I seem to still have the energy for edits and fixes (maybe even re-writing the whole thing). I just didn’t have the energy to get started.

There's a small (but growing) collection of meaningful tasks that I can now continue to work on even after my creative energy is spent for the day.

I have long subscribed to the notion that I have about 4 hours of high quality, "deep work" in me each day. Beyond that, I can do some meetings or admin tasks, but not much that shifts the needle.

Now I'm finding that even when my 4 hours are up and my creative energy is spent, I still have the energy to be productively prompting and overseeing AI-powered work.

I think is has the potential to shift the ideal knowledge worker's day like this (Your deep work to meetings-and-distracting-nonsense ratio may vary):

It’s strange because AI hasn’t really helped in other ways when it comes to my creative hours. When my energy levels are high, I much prefer still to write my work documents, my emails, my substacks etc.

The one big exception is coding. I only code for fun in the evenings and weekends, working on small side projects as an excuse to keep up to date with the latest AI.

It's something I enjoy, but I didn't have the energy for it after a day of work, until Gen-AI became good enough at coding.

In my day job, I can use AI to get me past the mental energy needed to get started on a doc or a deck, but it's not even fractionally as good at generic knowledge work as it is at vibe coding.

I am optimistic that my Google Workplace / Microsoft Office Suite can make big strides here over the next year. Even though the gains are unlikely to be as quick as they have been for code writing, any improvements have the potential to unlock additional hours in the day during which I can be productive.

The prospect makes me very excited.