I’m writing about 4,000 words a week at the moment, and keeping up the day job with Agentic consulting. Very often I think to myself, I’m not moving fast enough. I need to write more.
This newsletter is 750-1,000 words each week. Writing it takes 1-2 hours. I also post about AI every day on LinkedIn, with each post being about 200 words – so that’s another 1,400.
I don’t know where the time goes for that – I normally just put something into drafts whenever I have a good idea. The only structure I really have is that I have about 10 posts scheduled and move them around a lot. I barely ever write about new AI product releases – I think keeping on top of them would drive me crazy.
So probably about 2,000 words a week are published just from writing a weekly newsletter and publishing to LinkedIn every day. Perhaps I should try and get this time back via AI? I think about this a lot, but I barely use AI for writing at all.
Do you need AI assistance?
So much I see by people with big followings has some sort of AI assistance behind it now. Let’s take most other big newsletters in this space – like The Rundown or Future Tools – they’re generally AI generated roundups, but cleverly set up to look editorialised. In reality they’re on a kind of scraping autopilot by combining RSS and AI summarization. It takes a couple of hours to learn. It really is that easy. We don’t do it because LinkedIn too, apps like Saywhat and EasyGen enable you to write posts much more easily than coming up with it yourself. Barely a day goes by where I don’t see some ‘game changing’ new approach to content automation, be it through yet another Make workflow, or an agent that does everything for you. It’s like everyone wants to get away from the endless chore of content creation.
Enjoying the process
But I actually enjoy the creation part. I wouldn’t mind automating the occasional newsletter or LinkedIn post, but for the most part, I’m learning through the writing. It forces me to try and learn the hard stuff.
When we first started our AI in SEO white paper series (access all editions via the endnote), I was thinking, this should be pretty straight forward and we can just use AI – but the deeper I got into it, the more I could see that AI couldn’t do what we could.
It could do some Deep Research, but it was pretty unreliable. Indeed, I’d say our initial usage in February was downright disappointing. I got it to write some parts, but it could neither keep truly up to date with much accuracy, and moreover it could not form solid opinions.
For this reason I wouldn’t say we used it in a meaningful sense for writing. For research, yes – but not for writing. More than 2,000 people have requested access to our white papers so far.
Writing is for the benefit of the writer
This is where the other 2,000 words a week I currently write is going, in case you’re wondering. In direct conversation with some of those readers, they’ve commended the insights and writing. I’m glad of that because a lot goes into them. I don’t doubt that ChatGPT can write faster than us, and Deep Research can provide a lot of material very quickly, but neither of these things teaches us anything. That’s partly what writing is about.
I read in Andrew S Grove’s High Output Management that, at least in a business context, writing a document is often more beneficial to the writer than the reader. I usually write documents rather than create slide decks, and I’m inclined to agree.
For the reader, the concepts will often remain abstract, but the writer’s thoughts will be distilled. The document is a starting discussion point, which always needs clarification. Writing for internal business operations is often distilling thought, it is not ‘content’. We’re seeking to get things done and bring people along with that initiative, rather than purely inform (or entertain for some content).
In writing summaries, or curating the viewpoints or facts, then AI can do this well, although not cleanly and tonally right every time. It is an immense productivity tool in these areas. But by outsourcing specific thought to a robot, you’re taking an unnecessary risk by not distilling thought before action.
Worse still, you’re giving up the process of distilling your thoughts to AI.
That’s as true for writing a newsletter as it is for a business document. In choosing not to write, and in asking AI to do that for us, we are essentially removing a major thought process – and thinking of our content as a means for attention.’
Writing is for the writer’s benefit, not just for the readers.
