In February 2016 I set up a personal Slack workspace. No team. No collaborators. Just channels, organised by topic, where I could save things I wanted to remember. It was a scrapbook. A filing cabinet for the internet.
Ten years later, I ran a script across the full dataset: every channel, every save, every year from 2016 to 2026. I wanted to see what a decade of saving actually looked like.
The findings surprised me. Not because they revealed things I didn't know, but because they showed me exactly when things changed, often before I realised they had.
1. 2022 was a dead year
Almost every channel hit its lowest point simultaneously. #ai-artificial-intelligence: 7 saves, down from 41 the year before. #jobs: 23, down from 204. #strategy: 14. #food: 5. The collapse was across the board.
I don't have a clean explanation for it. Something happened in 2022. The data just shows a year where I stopped paying attention to nearly everything.
2. The AI inflection is a visible cliff
#ai-artificial-intelligence had 7 saves in 2022. ChatGPT launched in December of that year. In 2023: 581 saves. That's an 83x jump in a single year, the sharpest move in the entire dataset.
You can talk about inflection points in the abstract. This is what one actually looks like in personal data.
3. #cultural-interface didn't exist before 2023
It's the all-time number one channel by volume: 3,326 total saves. But it's only three years old. Created in 2023, immediately dominant, peaking at 1,253 saves in 2025 alone.
The thing that defines the workspace now didn't exist three years ago. That's worth sitting with.
4. #jobs never sleeps
Active every single year from 2016. Two massive spikes: 2021 with 204 saves (the post-pandemic job market) and 2024 with 435 saves, the biggest single-year figure of any channel in the entire workspace.
This channel tells the truth about where my head was, every year. It doesn't lie and it doesn't take breaks.
5. #strategy peaked in 2020 and never came back
172 saves in 2020. 134 in 2021. Then 14 in 2022, and it's been hovering below 61 ever since.
This is the shift from "strategist who studies the industry" to "founder who ships things," visible in a single channel's trajectory. I stopped collecting strategy frameworks when I started building products.
6. #food was a lockdown artefact
19 saves in 2018. 68 in 2019. Then 199 in 2020. Back to 69 in 2021 and collapse to 5 in 2022.
One year, one reason. A clean natural experiment in what happens to your interests when you can't leave the house.
7. #tools peaked in 2018 and has been declining for eight years
82 saves at peak. Now around 24 a year. When I was in agency world I collected other people's tools. As I started building my own things, the interest faded.
8. #events never missed a year
Active every single year from 2016 to 2026. Ten out of ten. From 2 saves in 2016 to a peak of 68 in 2023. Across every phase of my career, agency, Web3, AI, founder, this thread never broke.
It's the one consistent thing in the entire workspace.
9. #portfolio-page: nine years of the same question
Active every year since 2017, peaking at 92 in 2021. Never drops to zero. The question of how to present yourself professionally never goes away. It just changes shape.
I find this oddly reassuring. Everyone worries about their portfolio. Some of us just have nine years of evidence to prove it.
10. 2026 channels are all projects, not interests
Every new channel created in 2026 is named after something I'm building: #withmoshi-aivoice, #buggysmart, #claude-code, #pattern-media, #firstorder-aivoice. Zero "interest area" channels. The mental model has shifted entirely, from curation to production.
The bigger story
Run across all ten findings, the same pattern emerges. From 2016 to 2022, I was someone who studied and tracked things. Since 2023, I've been building them.
The channels show it before I could have said it. The decline in #strategy, the rise of #cultural-interface, the shift from tool-collecting to tool-making, the new channels all named after live products. None of this was planned. I didn't set out to document a transition. I was just saving links.
That's the thing about a decade of saves. It's not a bookmarking system. It's a mirror with a ten-year memory.
The analysis was built using a Python script pulling the full history from every channel via the Slack API. The interactive dashboard is at mikelitman.me/slack-dashboard.
