51) Continuing Education

Behind-the-scenes building Vambrace AI, a company on a mission to figure out its mission. Please pardon the stream-of-consciousness style. Subscribe to follow along or visit the site here:

(typos are to make sure you’re paying attention)

Introductory Remarks

Dear Vambracers —

In last week’s post, Validation, we explored the implications of OpenAI and Anthropic’s move into the consulting space. In general, it’s an encouraging sign and validates the opportunity around AI implementation. For us to ever truly enter an AI-first world, we will need to actively help organizations construct and operate AI-native systems that drive some business outcome in an efficient manner.

I also believe that entirely new systems for organizing and transforming information across organizations will be borne as we struggle with the on-the-ground challenges of building AI systems. And I’m excited to be on the frontlines of these difficult and exciting challenges.

Continuing Education

Background

To that end, I’ve been reflecting on the ways in which I work and assessing my competencies and interests and weaknesses, etc. As background, I really really love learning; college was one of my favorite times of life.

As I’ve been thinking about my day-to-day, so many things are coming at me so quickly from so many different angles that it can feel like I don’t have any quiet time to think and explore and learn new things. It can also feel like a luxury or an indulgence to try to develop new (or expand upon existing) skills when there are tasks on my plate.

But I’m realizing that continued learning is really something I love to do, and something that is vital to the energetic substance of my existence. So I’ve soft-committed to real continuing education in the form of skills development and little projects that I’m having AI guide me through. Specifically, I want to develop a more foundational understanding of core computer science principles (coding and algorithms), and I also want to thoughtfully improve the construction and quality of my writing.

I think it’s kind of funny that the two skills I’m interested in are believed to be directly threatened by AI. But the more I’m in the weeds on AI, the less concerned I am that it will replace true artists. I believe that mid-level output is at risk of automation, but folks who really truly understand some craft well enough to establish their own voice won’t be at risk, and their output will likely be amplified.

So as I continue to push to grow and improve as is required of me by my clients and as is dictated by my engagements, I want to be thoughtful about also revisiting the basics and trying to build a non-AI intuition around a lot of these things.

Programming

For programming, I’ve taken introductory courses in college and I can understand conceptually how data needs to flow throughout a system and stuff, but I have less of a firm grasp on what that actually means and how it happens. And then for more complex systems at the enterprise level, I think there are many more lower-level infrastructure-type considerations that I’ve started to scratch the surface on, but that I want to more technically understand.

I think AI is one of the most impactful technologies ever, but building with it pretty much 24/7 has been psychologically very confusing and challenging. I can write beautiful prompts, and then I can get into meta-prompting, and I can have Claude give Codex the most beautiful, prescriptive, and context-rich prompts ever—and I can leverage cutting edge adversarial prompting strategies—but really I’m going entirely off of vibes and whether the output functions correctly and the data is being properly stored throughout the backend. And that’s why it’s vibe-coding. I’d love to graduate from high-output (and lower-upper-middle-class quality) vibe-coding to beginner agentic engineering.

So I’m going to start challenging myself to learn and hand-type agentic loops and other relevant algorithms and stuff so I can actually start to understand and build intuition around what actually is going on here. And then I might be able to improve the quality and competence of my AI-generated code. From slop to pop!

Writing

For writing, I obviously enjoy the act of typing and I have a lot of thoughts that I like to share, but ever since I got a B in english freshman year of high school I’ve felt pretty inadequate about my relationship with the written word. I don’t really know the core tenets of sentence construction and narrative development and stuff, or like any tips and tricks or best practices, and so I’m revisiting the basics (starting with “what’s a noun”).

I’d love to structure my thoughts more clearly and feel more in control when I’m writing, and so that’s sort of the goal. Right now I’m just sort of going off of vibes and doing what feels right, but I know I write a lot of run-on sentences and probably have too many clauses like all the time; and then I also tend to just throw semicolons and hyphens in a lot for no real reason.

And then I’m sort there are hundreds of other hard- and soft-rules that I don’t follow. [Like I’m pretty sure you aren’t supposed to start sentences with “and.”] And I’ve never been great about really understanding active vs passive voice and stuff like that. I know enough just in terms of pattern-matching to put out mostly comprehensible prose, but I have never viewed it as a strength. [I also pretty much decide on when to make a new paragraph based on the length of the paragraph and the overall shape of the passage and how it looks on the page.]

I also hate proof-reading and editing and stuff—which is why I position this as a “stream-of-consciousness” newsletter, because in my mind that excuses me from tedious editing and, by saying it in the header, it makes you complicit in my grammatical incoherence and (in my mind) absolves me of your judgment.

Humility

One of the most difficult, but also most rewarding (hopefully), aspects of entrepreneurship is that it’s very humbling. Like I’m sort of slip-sliding my way through this. I make at least 3 mistakes a day (on a good day). And I go back and forth constantly between thinking I’m just a few years away from massive success and feeling like a complete and utter worthless failure.

The danger is that it can all become very gratuitous and self-involved and stuff (almost by definition). So then the pain and the #struggle can serve as a sort of drug. Meanwhile who knows when I’ll hit a real break, or if this is actual a constructive use of my time. And I could just be throwing away some of the best physical and cognitive years (maybe) of my life.

But the point I wanted to make here is that, given the above dynamics, it’s not a big thing to accept that I like learning and admit that I could use some training in the fundamentals of relevant skills.

Looking Forward

So, I’m metaphorically going back to school. One awesome use case of AI is that it can function as an all-knowing teacher in your pocket, and I intend to leverage that use-case to help me get some real learning under my belt. I’ll keep you posted on how it goes. Although I’d hope that you will notice a change in the quality and coherence of what you read here. But only time will tell.

Have a great week!

Sincerely,

Luke