The other day, I was sitting in front of my screen, looking at a comment under one of my posts. The post was about a founder who asked me which system he should choose - Notion or ClickUp. I looked at his current setup and told him neither would help, because what he already had in Notion was basically a junk drawer. And the tool wasn't the problem.
The post got some comments. And one of them stopped me cold.
Confident, well-written, full of smart terms like "headless workflow orchestration," "enforcing handoffs," and "expensive maintenance cemeteries." The point was: a single pane of glass in a Notion workspace is just a passive canvas that inevitably turns into a graveyard, where the team spends 15 hours a week updating pages nobody reads anyway. The real operational backbone is automated chains that control task handoffs, validate outputs, and sync state on their own. I read it. Then read it again. And... I didn't know how to respond.
Sometimes I think I'm a slow thinker. Not exactly the best confession for someone who advises other founders, right?
My first thought: damn, this is obvious, man, we're talking about different things. Second: wait, what made him write it this way? Maybe some painful experience? Third: I wanted to fire back with something like, come on, tell me about automation. I've been building this shit with my own hands for twenty years. I started typing. Deleted it. Started again. Deleted it again. Then leaned back in my chair, put my hands behind my head, and thought: why the hell can't I ever learn to respond quickly?
Funny, years ago I read a phrase in a book that stuck with me: rushing is only good when you have diarrhea. In every other case, it rarely leads anywhere good. Crude, but every time I feel that itch to answer or do something quickly, I remember it.
The thing is, I'd already seen this movie. I hadn't just "heard about automations." I lived in them. We tried every tool out there: Jira, Confluence, Trello, Airtable, ClickUp, Notion, Redmine. We built custom services and APIs. Of course, everything was documented in Jira... which became an even bigger junk drawer. Especially when suddenly everyone had to drop everything and figure out why a client's server had gone down. We used n8n when it had just appeared. Hell, I was so inspired by IDEO that when Linden Tibbets came up with his idea of digital duct tape that became IFTTT, I was one of its early users.
But… at one point, I even went back to Google Sheets. Simple and straightforward as a stick.
We jumped from one CRM to another - Zoho, then HubSpot, then Pipedrive - and every time wire them to Jira, or to our email system, or to our document management. Every migration consumed an enormous amount of time. And after every move, something stayed in the old system, something moved to the new one, and nobody really knew where the hell that contract or latest specs actually were.
We had an entire IT department that didn't work on client projects, only on maintaining our internal infrastructure. And a significant part of their time went into simply keeping all those integrations and automations running. Just keeping them alive. Of course, I didn't think about that back then. Definitely didn't calculate what all that time cost us. After all, we were building advanced systems. Systems that would watch over everything while we did cool creative work.
By the time we finally built a proper structure, we had almost forty people. And by then I'd realized something I stubbornly refused to see before: automation really starts working only when the process underneath it is stable and repeatable. When you're changing your approach every week, testing things, experimenting with brilliant new ideas - and at early stages there's no other way - every automation becomes yet another project you have to rebuild.
After sitting with it for a while, thinking it through, and remembering how we'd burned thousands of dollars under the banner "automate everything boring," I realized that on the main point, the commenter and I actually agree. System architecture comes first. And without proper maintenance, everything eventually becomes a graveyard. Just not only workspaces. Anything can become a graveyard. Automations rot just the same. Integrations break silently. Even the smartest AI agent in the world can go stale. The logic behind the architecture stops matching the processes it was built for. Someone has to notice, fix, change it. Someone alive.
Even now, building my own processes from scratch and being able to create almost any automation myself, I'm extremely conservative about it. Because I've fallen down that rabbit hole more than once — where you're constantly rebuilding something because something somewhere changed. Or, damn it, I have to admit - sometimes improving things is just fun. I've radically rebuilt my content creation process probably ten times in the last two years alone. If I'd followed every itch to build another agent or automation, I wouldn't be writing this letter now. I'd still be fixing or improving something.
For an agency of eight to fifteen people that's still figuring out the basics - who owns what, what does "done" look like, who makes which decisions - headless workflow orchestration isn't a solution. It's another problem disguised as progress. Because complexity looks serious. We've got a serious system here — must be a serious business. I genuinely used to think that way. But if your architecture requires a dedicated specialist and monthly payments just to stay alive, it isn't reducing friction. It's creating it.
So, after thinking a bit more, I decided that being a slow thinker isn't really a flaw. It's more like a professional deformation that, unfortunately, sometimes gets in the way in my personal life, but works surprisingly well in business architecture. I remember in university, when we were designing circuit boards, they taught us: think about where this component will be used. Will it overheat. How accessible will it be. How easy will it be to repair.
That slowness has saved me far more often than it has hurt me. From promises made without thinking. From words I'd regret later. From decisions whose consequences I'd have to live with. I prefer to weigh the full context before I respond. Even on consulting calls, I always let people know upfront that I'll take a couple of days and come back with a detailed breakdown. And when someone pushes for a quick answer, I always say the same thing: do you need to hear what you want to hear, or what can actually help?
Sure, I probably lose those who need everything urgent-urgent, fast-fast, go-go-go. But that doesn't bother me. I'd rather work with people who want to build something designed to survive time and the punches reality throws at it. So the first question I ask now isn't "what sophisticated system can we build?" It's: what's the simplest architecture this specific business can actually own and maintain when I'm not in the room?
Look at your current stack of tools and automations. Ask yourself: if I disappeared for two weeks, could someone on my team understand how all of this works and fix it if something broke? If the answer is "no," maybe your architecture is serving whatever is possible, just not your business.
So, I wonder, when you want to bolt on the next shiny thing — are you solving a real problem? Or do you just rush to look like a serious business? While you are still going to bed anxious that half of tomorrow will be spent remembering where things are and how they work.
P.S. I did reply to that comment. A day later. Calmly, to the point, and with how I actually see it: every stage of a business needs its own architecture. The answer was more precise and thoughtful than any of the ones I started typing and deleting in the first ten minutes. Turns out, sometimes a day of silence is worth more than five minutes of wit.



