Everything Is on the Table
May 2026
I spent twenty hours writing What Are Humans to Do After AI?. I spent twenty hours staring at the skylights in Dandelion Chocolate and walking down the streets of the Mission, picturing what knowledge work will look like a decade from now. After writing and rewriting, I finally called my essay done and laid it to rest. But the ideas kept burning in the back of my mind.
I found myself incredibly impatient at my day job. I work as a software engineer at a company that builds infrastructure for analyzing company data. My projects to help companies process millions of documents felt intellectually stimulating, but what about this future of knowledge work I so vividly imagined?
And so, I went home and started a doc titled: The Place to Do All Knowledge Work.
The doc was simple. I first wrote out what questions a marketing analyst, data engineer, research scientist focus on in their jobs today and what I hoped they would ask if they had more powerful tools. Then I asked myself:
“What does the tool look like?”
“What technologies and resources have to exist to create this tool?”
“Which ones does my company excel at? Which ones do we need to work on?”
“What is an end to end workflow that, if worked, proves that my product and technology hypotheses are correct?”
“What’s the best sequencing for what to work on? What can I start doing today?”
The answer to that last question is where I began. From my doc, I concluded that there was a set of workflows that my company’s products could not do today, and if we invested in X, we would be a big step closer to the fully autonomous task execution I imagined.
I started sending my doc to my teammates, and asked if they knew anyone relevant. I went on walks and got on video calls with every tech lead referred to me. I walked them through how I concluded to build X, listened to their reactions, and asked for artifacts to help me build context. After hours, I used coding agents to work through design docs. I took the code pointers I was given, ran each team’s product and evaluation benchmarks locally, and played around with the product. On my commute home, I read papers and looked at how other companies approached the problems I started to notice. After about seven conversations and twenty hours mining our docs and codebase, I had a full landscape of what each team had built, how their products worked, how they performed, and why we still couldn’t accomplish the workflows I wanted to see. From there, I identified what I believed we needed to invest in next and split my time between my existing projects and my proposed one.
In the midst of my side quest, I found myself in the office at 10pm. The floor was silent. The ceiling lights had turned off. Alone, I looked past my monitor at the San Francisco skyline, grinning like a maniac. I felt a chaotic sense of power. A feeling of being unstoppable.
I felt like everything is on the table.
Our product home screen. I find it to be overwhelming. A new user with a task in mind wouldn’t know where to start. Right now, I can screenshot our home screen, throw it into a coding agent, discuss why it feels overwhelming, find the most recent contributors, propose my updates, and upon alignment, ship the change that helps people get started faster.
Our data science agent. I asked it to edit some dashboards and it couldn’t figure out how to speed up the loading time of my queries. Right now, I can use a coding agent to find the data science agent skills, do some prompt engineering, add new test cases to the relevant evals, verify my changes, and contribute code that makes the product faster for everyone.
My company has twelve thousand employees and at least thirty products. We have many processes, legal, security, scheduled releases, and a heck ton of code. I can choose to feel like a shrimp in an unmovable ocean. Or, I can choose to believe that everything I notice matters, that my influence on our product is as great as my ability to recognize the things I can touch.
But believing this is only half the picture.
Today, people love talking about being agentic. We point out that AI helps us learn quickly. That now, we can just do things.
But, when doing something is as easy as a natural language prompt, I fear that more things will be done, but not many more things will be done well. There is a difference between “I have a spark of inspiration, let me vibe code it today” and “there is something I want people to experience, I feel opinionated about it, and now I know why I’m going to take the next step”.
I started writing What Are Humans to Do After AI? because I disliked the hopeless AI doomer energy that permeated my conversations with my friends, and I felt confused why I didn’t agree with them. And so, I sat with myself in Dandelion Chocolate, for hours and hours thinking about how knowledge work is changing. Every sentence I edited and re-edited made it clearer to me what I wanted my future job to feel more or less like. I developed a bar for what I want people to experience. My desire to reach that bar and my distaste toward an unthoughtful alternative gave me the energy to write The Place to Do All Knowledge Work after a long day in the office.
There are so many things you can just do. There are only so many things you can care about. Everything is on the table. All that matters now is what you care to change.
If you made it this far, thank you for reading my May essay. If you have questions, opinions, experiences about any of these topics, I would love to discuss! That’s why I write after all :) You can find me @jjanezhang on X.
A special shoutout to my friends Jessica, Blake, Jacob, Nathan, Amy, and Andrew!
Other topics I thought about but didn’t write about:
1.
Protecting people’s brains. I have this funny imagery in my head after I converse with someone and glean that they have a very cool, daring, optimistic view of the world. I imagine taking their brain out of their head and holding it the way a young girl holds a rabbit in her arms. I feel that some people have incredible world views, and I almost want to physically protect them, so as to prevent them from becoming jaded.
2.
I don’t want my phone. I’m sitting on the airplane waiting to take off, and from my aisle seat view, all I see is a column of thumbs swiping on screen after screen. The person directly to my left is flipping through lipstick color swatches. The person in front of her is watching some influencer on Instagram. All I want to do is go back to early 2000s and live a day when no one had a smart phone, and we just sat on the plane with ourselves.
3.
Wear a wig. Put on a temporary tattoo. A good friend of mine recently turned 40 and threw an 80s themed birthday party where he required everyone to wear a wig and to put on some very large Americana tattoos. I didn’t know most of the people, but sitting next to someone wearing a neon orange leotard or a full chest and sleeve of temporary tats makes me feel like I’m just another human next to another human.



Aw I love the humanization of Blake's birthday party protocol! Also love your insistent drive to challenge the status quo of your company's products.