Things I’m wrestling with

I’m trying something different this week.

I’m writing this whole newsletter as a brain dump using Wispr Flow to capture as I just talk. I’m literally just talking and it’s writing it out as I speak.

There’s a lot of stuff I’ve been pondering and I’m just dumping here. I don’t have it figured out and actively trying to find a path through.

In a world being eaten alive by AI, I think we all need more of a real, raw, and unpolished look behind the scenes to remind us that we’re human and we’re all figuring it out. Hopefully you find it useful, but if not let me know. I genuinely want this newsletter to be something you find valuable and isn’t just another email you archive or delete or ignore.

1. What does an AI native team actually look like in 2026 and beyond?

I've been thinking a lot about this and where I'm netting out is you need a strategist and a builder partnered together at the hip to deliver value. All they are thinking about is solving problems and figuring out where the value is.

The strategist is owning the business side. They’re figuring out what's going to move the business forward and what customers need and want. The builder is understanding and owning the technical requirements and orchestrating agents to build.

They should be doing this together with pretty blurry roles. They should both be equal partners in the solution and it should be highly collaborative. The work should be more about the nuance, cultivating relationships, finding unexpected intersections, and zagging while AI is zigging.

2. How do you actually incentivize people properly in a world where everything is being done by AI?

What are the right metrics to track? What are the right ways to frame success?

Currently I think we should be focusing on inputs rather than outputs. People should be using tools to do their work. They should be focusing on relationships and actual impact and value. They should be thinking strategically about what problem is the right problem to solve, things like that.

What's hard is mapping this kind of thing and figuring out what are the correct leading indicators and then how to actually incentivize people to do it.

3. What's the right blend and balance of people and AI?

How much should you really automate or have handled by agentic workflows?

I've been thinking a lot about this whole concept of an autonomous business, basically building a business that can run in large part without having to do a lot of the operational overhead that normally comes with running a business. The difficult part is really evaluating how much to actually give up to AI. At what point does that change the perceived value? Also at what point does the perception curve change where everyone expects it and actually they want the thing to be done fast and they want it to be done cheap and the only way to accomplish that is with AI tools and the perceived value isn't lower or different.

4. What are the characteristics of a worthy problem that's actually worth trying to solve?

It's really interesting to think about this because right now with AI it's just so easy to solve problems. You can basically describe to Claude or ChatGPT or Gemini whatever the situation or context is and you can provide it with access to tools. You can do things to make it so that it can essentially solve that problem for you especially when we're talking about administrative stuff, operational stuff, software stuff. Those things become way easier to do. So then it becomes a question of, is this really the right problem to solve? Is this a worthy problem to solve? I think that is really hard sometimes to decide what is really a worthy problem.

5. How do I not be the bottleneck for AI?

I think about this a lot because often I'm having to remind myself, “What was I working on? What was I having AI work on? Where were we in the process of doing yesterday? How do I use the things we just built?

My capacity is constraining AI's capacity. To give a real example, I've now got my entire Claude Code set up on a virtual machine and I'm able to access it through an SSH app called Termius on my iPhone. What this means is I can now run a bunch of stuff independent of my computer and have full access to a computer and all the different tools and things I'm using to help me run StealthX actually efficiently. What's funny is as I've been building all this stuff out, I've been having a hard time just remembering where we left off. I have to constantly keep asking it, "Hey, what were we doing? Remind me where we were." Obviously AI is really good at reminding and keeping track of everything, creating session logs and saving memories and all this kind of stuff, but I get lost in the sauce sometimes and forget where we were and what we were doing. I’m constantly pondering what the best way of working with AI is. What’s the best mental model? What’s the best way for me to not slow AI down?

6. How do you actually do multiplayer?

I don't think anyone's really cracked the code on multiplayer with AI. Everyone's still building their own thing that's making their life better, making their job easier, but when it comes to sharing with others, it's really complicated because there's permissions, roles, and scopes. You don't really want everyone on your team to have access to everything because then it could end up having exposure to something that they shouldn't, like compensation data or things like that.

Those are the kinds of things that it's hard to work through. There are obviously tried and true engineering principles that you can leverage to help you architect that, but it's still complicated to think through all the where the rails need to be and how they need to be laid out.

Also as fast as everything's changing, I wonder “Am I wasting my time and energy building this out when I could just use something off the shelf in two months?” Tough to discern the best path to effective multiplayer right now.

Wrapping up

Anyways these are some things I’m continually pondering and I haven't really figured out the best approach. AI is very Wild West and there’s not really a proven “Oh, yeah this is the way” yet.

I'm very curious to hear if anybody else has been tackling it or has thoughts and opinions. I'd love to chat with you and learn from each other.

If nothing else I hope this newsletter makes you feel better about any questions that you're having and that you're not alone in wondering these kinds of things.

Hope everybody had a wonderful 4th of July!

Onward & upward,
Drew