I just interviewed the Director of Engineering at X about content, AI's impact on writing, and the rise of the generalist freelancer, check it out →​

​

​

Here's my full visualized notes from the interview (lemme know what you think of this format)!

0:00: Daniel sold his company Laskie to Elon Musk at Twitter/X where he is now a director of engineering building out the X hiring platform and other monetization products. He also sold a YC-backed hiring platform to Indeed and is currently creating an incubator for AI engineers.

⬇

​

2:40: He built a company called Laskie which was bought by Elon Musk directly when he acquired X.

⬇

4:10: His co-founder was just posting on Twitter about their company, Elon Musk read them, and he DM’d him asking to do the same product for Twitter.

⬇

6:57: Their first meeting with Elon was at his birthday party in Palo Alto with jugglers and fire breathers, and they sat down for 45 minutes during the party going over their vision for the X hiring product. They never even got past their first slide of their 20-slide deck, and he made an offer on the spot.

⬇

8:20: They made a 20-slide deck, but in reality only got to a slide about how much money LinkedIn and Indeed make and never made it past that. Instead, they chatted about more deep and technical details the whole time not suited for slides.

⬇

10:50: During price negotiation, Elon just threw out a price on the spot. The deal even got put on hold for about 2 months but it was during the Twitter-to-X transition and then it happened right away after the dust settled.

⬇

13:27: Why try to bring hiring to X? LinkedIn has more pandering content, whereas X is far more hardcore. This can be good for indie hackers and designers, but maybe not great for buttoned-up lawyers.

⬇

15:30: When hiring a designer (for example) you don’t care about their resume, more just their work. With X, it’s easy to see.

⬇

​

21:35: Daniel has all these beta features enabled on his account most users don’t have like expanded bios, featured job posts, jobs search.

⬇

​

25:50: I’m seeing the world of freelancing and social media converging; it’s currently very efficient. X might be working on something that can help with this.

⬇

​

29:20: UpWork has a monopoly on a lot of freelancers, but the platform STINKS. I did an experiment on UpWork to test it out, and the usability is absolute utter trash. The only place that’s done well at competing is Fiverr, which is very simple to use. Daniel’s startup Laskie was basically an UpWork and Fiverr clone.

⬇

​

32:18: How do you think about new features at X? You can’t just throw traditional recruiting stuff at X, they are working on different ways of combining job search with your feed. Or maybe showing you the posts of people looking for work or hiring. They want better DMs also, you need to be able to have 1,000’s of DM conversations and the current system can’t do it. Basic product UX issues are coming first.

⬇

​

37:48: Making X a “super app” that does everything. Much like WeChat in China. That seems to be the end roadmap. Two things Elon says all the time are X is “The Internet Town Square” and should be “The Everything App.” WeChat was originally a social network and messaging app, then got into payments, then ballooned from there. WeChat has a billion+ Daily Active Users and is an integral part of people’s lives.

⬇

​

40:00: During “The Internet Bubble” around 2000 a lot of the ideas were good, but they were too early….is it the same with AI right now? Daniel sees AI just starting, so yes, there’s too much money being invested in not-useful ideas (or will be done for free by more advanced models). Daniel sees 5 levels of AI: 1.) ChatBot 2.) Reasoner 3.) AI Agents 4.) Creative problem solver 5.) Organization on its own. He uses Claude and Cursor to be a 10x more effective programmer. AI Agents might currently be in a bubble, but eventually that technology can likely run entire organizations and actually manage humans.

⬇

​

46:46: Chat GPT is barely 2 years old. The evolution from GPT-2 to the current model is unreal, and went from “novel” to “super useful.”

⬇

​

47:28: I run a writing company, and I’ve seen AI wipe out several types of writing, and morph what writing has become. 3 types of writing: Low Level, Medium Level, and High Level Writing. For example, AI can write decent blog posts, but most people aren’t reading those anymore. Many companies moved to 100% AI writing, but now are bringing the humans back.

⬇

50:18: Daniel thinks there will need to be a human in the loop to see if other humans care. Neville says, “I want my X feed to be human, I want my customer support to be AI.”

⬇

​

52:00: AI writing demos are crazy cool, but when you really use it, it’s too generic. However, it’s super good for research.

⬇

​

53:18: Daniel will write by collecting a lot of links and write a lot of thoughts, then have AI compile the insights of all those. It’s not writing “for” him but rather “with” him.

⬇

54:31: AI takes your job and moves you “up the chain” to be more powerful. For example, if you can’t do design well, it can help you do that. Learning these tools can make you a “Triple Threat” much faster.

⬇

​

57:20: Since the 1960s, many writers have moved “up the chain” into business owners thanks to better technology.

⬇

​

58:10: Daniel thinks there’s going to be a “rise of the generalist” with these new AI tools because you can now be good at writing, programming, planning, etc., with AI help. You can pretty much be a team of one. It can help you do projects you couldn’t do before because AI will help you fill in your skill gaps.

⬇

1:00:18: A guy in our Office Hours built a #1 ranking calculator, we thought he hired a programmer, but he just asked ChatGPT to program it. He “moved up the chain” in his job with help from AI.

⬇

​

1:04:18: What Daniel is building in AI. He’s gone into hardcore experimentation phase by building a bunch of AI tools quickly. He built a writing drafting tool, a personal Yelp replacement tool that tells you what fun things to do based on your preferences.

⬇

1:08:05: How do you overcome the problem of the next versions of ChatGPT just doing what your company does? You must really understand what the end user is trying to do.

⬇

​

1:11:07: What new business would you start again today? I would start looking at 2nd and 3rd order effects of AI and see what impacts each will have. Figuring out curation in an AI world. Looking at what it took big teams to make, but could I make it with a small team.

⬇

1:13:15: Career advice in AI world? Increase your luck surface area. Putting yourself out into the world so serendipity can happen. Use to learn AI tools and be a common generalist.

⬇

​

1:15:00: Your direct report is Elon Musk, any learnings? He has an unbelievable sense of urgency on himself and teams. He needed to move a data center and just did it himself and some employees on Christmas.

⬇

1:18:00: Is the controversy he causes a bug or a feature? He can shoot himself in the foot, but on the other hand, he’s 100% himself and who you can expect. Steve Jobs was similar, but in a way, it might actually be more efficient. Steve Jobs said, “Blunt is simple.” People know where they stand instantly. Elon also has insane endurance, like constant input and output all day long. He’s able to context switch instantly and have in-depth knowledge on each subject.

⬇

​

1:24:48: Daniel is going to do a free AI Accelerator for engineers who want to build things and ship them quickly. Doing this just to meet other cracked engineers.

⬇

Watch the whole interview here (leave a comment if you enjoy these)!

​

​

Follow Daniel O'shea:​
- x.com/oshea​
- x.com/jobs​
- Daniel’s AI Accelerator: Timetobuild.dev

​

Hope you enjoyed this interview!​
​Sincerely,
Neville Medhora

​​​​​​​

​

​