I remember cringing at this one line from an episode of the 1990s TV series, Star Trek: Voyager:
Computer, install a recursive algorithm!
I always thought that you would never program a computer that way…until now.
Category: Artificial Intelligence
Here’s a promising start to the new year: thanks to a successful appearance on the Intelligent Machines podcast back in October, I was a guest on episode 1065 of Leo Laporte’s main podcast, This Week in Tech.
Leo, Blackbird.AI’s Dan Patterson, and I spent just under three hours on Sunday talking about the week’s tech news and having fun while doing so. The episode takes its title, AI Action Park, from Action Park, an insanely dangerous theme park that I mentioned while we were talking about DeepSeek’s Manifold-Constrained Hyper-Connections architecture.
You’ve probably seen this tweet, written by none other than Andrej Karpathy, founding member of OpenAI, former director of AI at Tesla, and creator of the Zero to Hero video tutorial series on AI development from first principles:
I’ve never felt this much behind as a programmer. The profession is being dramatically refactored as the bits contributed by the programmer are increasingly sparse and between. I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like skill issue. There’s a new programmable layer of abstraction to master (in addition to the usual layers below) involving agents, subagents, their prompts, contexts, memory, modes, permissions, tools, plugins, skills, hooks, MCP, LSP, slash commands, workflows, IDE integrations, and a need to build an all-encompassing mental model for strengths and pitfalls of fundamentally stochastic, fallible, unintelligible and changing entities suddenly intermingled with what used to be good old fashioned engineering. Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.
It would be perfectly natural to react to this tweet like so:
After all, this “falling behind” statement isn’t coming from just any programmer, but a programmer who’s been so far ahead of most of us for so long that he’s the one who coined the term vibe coding in the first place — and the term’s first anniversary isn’t unit next month:
Karpathy’s post came at the end of 2025, so I thought I’d share my thoughts on it — in the form of a “battle plan” for how I’m going to approach AI in 2026.
It has eight parts, listed below:
- Accept “falling behind” as the new normal. Learn to live with it and work around it.
- Understand that our sense of time has been altered by recent events.
- Forget “mastery.” Go for continuous, lightweight experimentation instead.
- Less coding, more developing.
- Yes, AI is an “alien tool,” but what if that alien tool is a droid instead of a probe?
- Other ideas I’m still working out
- This may be the “new normal” for Karpathy, but it’s just the “same old normal” for my dumb ass.
- The difference between an adventure and an ordeal is attitude.
1. Accept “falling behind” as the new normal. Learn to live with it and work around it.
Even before the current age of AI, tech was already moving at a pretty frantic pace, and it was pretty hard to keep up. That’s why we make jokes like the slide pictured above, or the Pokemon or Big Data? quiz.
As a result, many people make the choice between “going deep” and specializing in a few things or “going wide” and being a generalist with a little knowledge over many areas. While the approaches are quite different, they have one thing in common: they’re built on the acceptance that you can’t be skilled at everything.
With that in mind, let me present you with an idea that might seem uncomfortable to some of you: Accept “falling behind” as the new normal. Learn to live with it and work around it.
A lot of developers, myself included, accepted being “behind” as our normal state of affairs for years. We’re still in the stone ages of our field — the definition of “computable” won’t even be 100 years old until the next decade — so we should expect changes to continue to come at a fast and furious pace during our lifetimes.
Don’t think of “being behind” as being a personal failing, but as a sensible, sanity-preserving way of looking at the tech world.
It’s a sensible approach to the world that Karpathy describes, which is a firehose of agents, prompts, and stochastic systems, all of which lack established best practices, mature frameworks, or even documentation. If you’re feeling “current” and “on top of things” in the current AI era, it means you don’t understand the situation.
That feeling of playing perpetual “catch-up?” That’s proof you are actively playing on the new frontier, where the map is getting redrawn every day. It means you’ve got a mindset suited for the current Age of AI.
2. Understand that our sense of time has been altered by recent events.
I think some of Karpathy’s feeling comes from how the pandemic and lockdowns messed up our sense of time. Events from years ago feel like they just happened, and events from months in the past feel like a lifetime ago.
AI — and once again, I’m talking about AI after ChatGPT — sometimes feels like it’s been around for a long time, but it’s still a recent development.
“How recent?” you might ask.
Think of it this way: the non-D&D playing world was introduced to Vecna through season 4 of Stranger Things months before it was introduced to ChatGPT.
Need more perspective? Here are more things that “just happened” that also predate the present AI age:
- The Russia/Ukraine War started in February 2022, over 9 months before ChatGPT.
- Elon Musk made his $44 billion offer to buy Twitter in April 2022, tried to back out in July, and, after being forced by court intervention, bought it in October, which means all that drama happened before ChatGPT.
- The FBI executed their search warrant of Donald Trump’s Mar-a-Lago residence in August 2022, finding seizing of over at least 300 classified government documents and 48 empty folders labeled “classified” — 3 months before ChatGPT.
- These superhero movies were released in 2022, and all of them predate ChatGPT:
- Doctor Strange in the Multiverse of Madness (May)
- Thor: Love and Thunder (June)
- Black Panther: Wakanda Forever (October)
- Black Adam (October)
- The Will Smith / Chris Rock slap happened in April — 7 months before ChatGPT.
(Just recalling about “the slap” made me think “Wow, that was a while back.” In fact, I get the feeling that the only person who remembers it as if it happened yesterday is Chris Rock.)
All these examples are pretty recent news, and they all happened before that fateful day — November 30, 2022 — when ChatGPT was unleashed on an unsuspecting world.
3. Forget “mastery.” Go for continuous, lightweight experimentation instead.
Accepting “behind” as the new normal turns any anxiety you may be feeling into a strategic advantage. It changes your mindset to one where you embrace continuous, lightweight experimentation rather than mastery. You know, that “growth mindset” thing that Carol Dweick keeps going on about.
Put your energy into the skill of learning and critically evaluating new basic tools and skills that are subject to change over the gathering static domain knowledge that you think will be timeless.
We’re emerging from an older era where code was scarce and expensive. It used to take time and effort (and as a result, money) to produce code, which is why a lot of software engineering is based on the concept of code reuse and why older-school OOP developers are hung up on the concept of inheritance. Now that AI can generate screens of code in a snap, we’re going to need to change the way we do development.
My 2026 developer strategy will roughly follow these steps:
- Embracing ephemeral code: I’m adopting the mindset of “post code-scarcity” or “code abundance.” I’ll happily fire up
$TOOL_OF_THE_MOMENTand have it generate lots of lines of code that I won’t mind deleting after I’ve gotten what I need out of it. The idea is to drive the “cost” of experimentation down to zero, which means I’ll do more experimenting. - Try new things, constantly, but not all at once: My plan is to dedicate a week or two to one thing and experiment with it. Examples:
- First week or so: Prompts, especially going beyond basic instructions. Play with few-shot prompting, chain-of-thought, and providing context. Look at r/ChatGPTPromptGenius/ and similar places for ideas.
- Following week or so: Agents. Build a simple agent using a guide or framework. Understand its core components: reasoning, tools, and memory.
- Week or so after that: Tools and integrations. Give an agent the ability to search the web, call an API, or write to a file.
- Learn by teaching and building: Active learning is the most efficient learning! Building something and then showing others how to build it is my go-to trick for getting good at some aspect of tech, and it’s why this blog exists, why the Tampa Bay Tech Events List exists, why the Global Nerdy YouTube channel exists, and why I co-organize the Tampa Bay AI Meetup and Tampa Bay Python.
4. Less coding, more developing.
I used to laugh at this scene from Star Trek: Voyager, but damn, it’s pretty close to what we can do now…
Your enduring value as a developer or techies is going to move “up the stack,” away from remembering the minutae API call parameter order and syntax and towards “big picture” things like system design, architectural judgment, and the critical oversight, judgement, and quality control required to pull together AI components that are stochastic and fundamentally unpredictable.
That “behind” feeling? That’s the necessary friction for keeping your footing while climbing the much taller ladder that development. Your expertise is less about for loops and design patterns more about system design, problem decomposition, and even taste. You’re more focused on solving the users’ problems (and therefore, operating closer to the user) and providing what the AI can’t.
Worry less about specific tools, and more about principles. Specific tools — LangChain and LlamaIndex, I’m lookin’ right at you — will change rapidly. They may be drastically different or even replaced by something else this time next year! (Maybe next month!) Focus on understanding the underlying principles of agentic reasoning, prompt engineering, and (ugh, this term drives me crazy, and I can’t put my finger on why) — “workflow orchestration.”
Programming isn’t being replaced. It’s being refactored. The developer’s role is becoming one of a conductor or integrator, writing sparse “glue code” to orchestrate powerful, alien AI components. (Come to think of it, that’s not all too different from what we were doing before; there’s just an additional layer of abstraction now.)
5. Yes, AI is an “alien tool,” but what if that alien tool is a droid instead of a probe?
Let’s take a closer look at the last two lines of Karpathy’s tweet:
Clearly some powerful alien tool was handed around except it comes with no manual and everyone has to figure out how to hold it and operate it, while the resulting magnitude 9 earthquake is rocking the profession. Roll up your sleeves to not fall behind.
First, let me say that I think Karpathy’s choice of “alien tool” as a metaphor is at least a little bit colored by Silicon Valley mentality of disruption — that tech is for doing things to fields, industries, or people instead of for them. There’s also the popular culture portrayal of alien tools, and I think he feels he’s been getting the “alien tools doing things to me” feeling lately:
But what if that “alien tool” is something that can we can converse with? What if we reframe “alien too” to…“droid?”
Unlike an interpreter or compiler that gives one of those classic cryptic error messages or a framework or library with fixed documentation, you can actively interrogate AI, just like you can interrogate a Star Wars droid (or even the Millennium Falcon’s computer, which is a collection of droids). When an AI generates code that you can’t make sense of, you’re not left to reverse-engineer its logic unassisted. You can demand an explanation in plain language. You ask it to walk through its solution step by step or challenge its choices (and better yet, do it Samuel L. Jackson style):
This transforms debugging and learning from a solitary puzzle into a dialogue. This is the real-life version of droids in Star Wars and computers in Star Trek! Ask whatever AI tool you’re using to refactor its code for readability, make it give you the “Explain it is if I’m a junior dev” walkthrough of a complex algorithm, or debate the trade-offs between two architectural approaches. Turn AI into a tireless, on-demand pair programmer and tutor!
By reframing AI not as an alien tool, but as a Star Wars droid (or if you prefer, Star Trek computer), you can change the pace at which you can understand and manage systems. Unfamiliar libraries and cryptic errors are no longer major show-stoppers, but speed bumps that you can overcome with a Socratic dialogue to build your understanding. The AI-as-droid approach allows you to rapidly decompose and reconstruct the AI’s own output, turning its stochastic suggestions into knowledge and understanding that you can carry forward.
In the end, you’re moving from merely accepting or rejecting its code to using conversation to get clarity, both in the AI’s output and in your own mental model. By treating AI not as an alien probe but as a droid, the “alien tool” becomes less alien through dialogue. The terra incognita of this new age won’t be navigated by a map, but by directed exploration with the assistance of a local guide you can question at every turn.
6. Other ideas I’m still working out
Like “Todd” from BoJack Horseman, I’m still working out some ideas. I’ve listed them here so that you can get an advance look at them; I expect to cover them in upcoming videos on the Global Nerdy YouTube channel.
These ideas, taken together, are a call not to blindly climb the AI hype curve, but a call to develop a sophisticated, expert-level understanding of its limits. I hope to make them the basis of a structured way to conduct that research without falling into the “Mount Dumbass” of overconfidence.
-
Your tech / developer expertise is the guardrail. I believe that knowledge is the antidote to AI’s Dunning-Kruger effect. What you bring to the table in the Age of AI is critical evaluation, debugging, and architectural oversight. AI generates candidates; you approve or reject them, or, to quote Nick Fury…
-
Adopt a skeptical, experimental stance. Let’s follow Karpathy’s own method: try the new tools on non-critical projects. When they fail (as he said they did for
nanochat), analyze why they failed. This hands-on experience with failure builds the accurate mental model he described as lacking. -
Focus on understanding AI “psychology.” Understand the new stochastic layer provided by AI not to worship it, but to debug it (please stop treating AI like a god). Learn about prompts, context windows, and agent frameworks so you can diagnose why an AI produces bad code or an agent gets stuck. This turns a weakness into a diagnosable system.
-
Prioritize team and talent dynamics: You’ll hear and read losts of stories and articles warning of talent leaving as a result of AI. If you’re in a leadership or decision-making role, focus on creating an environment where critical thinking about tools is valued over blind adoption. Trust your team, and protect their “state of flow” and their deep work.
7. This may be the “new normal” for Karpathy, but it’s just the “same old normal” for my dumb ass.
Maybe it’s because he’s made some really amazing stuff that he’s surprised that the wave of change that it brought about has come back to bite him. For most of the rest of us — once again, that includes me — we’ve always been trying our level best to keep up, and doing what we can to manage our tiny corners of the tech world.
The take-away here is that if the guy who helped make vibe coding a reality and coined the term “vibe coding” is feeling a bit overwhelmed, we can take comfort that we’re not alone. Welcome to our club, Andrej!
8. The difference between an adventure and an ordeal is attitude.
Yes, it’s a lot. Yes, I’m overwhelmed. Yes, I’m trying to catch up.
But it’s also exciting. It’s a whole new world. It’s full of possibilities.
I’m outside my comfort zone, but that’s where the magic happens.
Here’s your start-of-the-year reminder that you don’t have to accept your LLM’s answers as gospel. In fact, you do what I do — talk back to them Samuel L. Jackson / Nick Fury from The Avengers-style.
The screenshot above comes from an exchange I had with Gemini earlier today. I like using LLMs as a sounding board for ideas — as I like to say “the one thing you can’t do, no matter how creative or clever you are, is come up with ideas you’d never think of.”
Gemini suggested a course of action that I completely disagreed with, so I decided to respond with one of my favorite lines from the first Avengers film, and it responded with “touché.” Keep thinking, and don’t completely outsource your brain to AI!
And just as a treat, here’s that scene from The Avengers:
https://www.youtube.com/shorts/NPXjPWIj31g
A lot of the drudgery behind assembling the “Tampa Bay Tech Events” list I post on this blog every week is done by a Jupyter Notebook that I started a few years ago and which I tweak every couple of months. I built it to turn a manual task that once took the better part of my Saturday afternoons into a (largely) automated exercise that takes no more than half an hour.
The latest improvement was the addition of AI to help with the process of deciding whether or not to include an event in the list.
In the Notebook, there’s one script creates a new post in Global Nerdy’s WordPress, complete with title and “boilerplate” content that appears in every edition of the Tech Events list.
Then I run the script that scrapes Meetup.com for tech events that are scheduled for a specific day. That script generates a checklist like the one pictured below. I review the list and check any event that I think belongs in the list and uncheck any event that I think doesn’t belong:
In the previous version of the Notebook, all events in the checklist were checked by default. I would uncheck any event that I thought didn’t belong in the list, such as one for real estate developers instead of software developers, as well as events that seemed more like lead generation disguised as a meetup.
The new AI-assisted version of the Notebook uses an LLM to review the description of each event and assign a 0 – 100 relevance score and the rationale for that score to that event. Any event with a score of 50 or higher is checked, and anything with a score below 50 is unchecked. The Notebook displays the score in the checklist, and I can click on the “disclosure triangle” beside that score to see the rationale or a link to view the event’s Meetup page.
In the screenshot below, I’ve clicked on the disclosure triangle for the Toastmasters District 48 meetup score (75) to see what the rationale for that score was:
For contrast, consider the screenshot below, where I’ve clicked on the disclosure triangle for Tampa LevelUp Events: Breakthrough emotional eating with Hyponotherapy. Its score is 0, and clicking on the triangle displays the rationale for that score:
One more example! Here’s Tea Tavern Dungeons and Dragons Meetup Group, whose score is 85, along with that score’s rationale:
I don’t always accept the judgement of the LLM. For example, it assigned a relevance score of 40 to Bitcoiners of Southern Florida:
Those of you who know me know how I feel about cryptocurrency…
…but there are a lot of techies who are into it, so I check the less-scammy Bitcoin meetups despite their low scores (there are questionable ones that I leave unchecked). I’ll have to update the prompt for the LLM to include certain Bitcoin events.
Speaking of prompts, here’s the cell in the Notebook where I define the function that calls the LLM to rate events based on their descriptions. You’ll see the prompt that gets sent to the LLM, along with the specific LLM I’m using: DeepSeek!
So far, I’m getting good results from DeepSeek. I’m also getting good savings by using it as opposed to OpenAI or Claude. To rate a week’s worth of events, it costs me a couple of pennies with DeepSeek, as opposed to a couple of dollars with OpenAI or Claude. Since I don’t make any money from publishing the list, I’ve got to go with the least expensive option.
This meme’s going in an upcoming video on the Global Nerdy YouTube channel.
If you watch just one AI video before Christmas, make it lecture 9 from AI pioneer Andrew Ng’s CS230 class at Stanford, which is a brutally honest playbook for navigating a career in Artificial Intelligence.
You can watch the video of the lecture on YouTube.
The class starts with Ng sharing some of his thoughts about the AI job market before handing the reins over to guest speaker Laurence Moroney, Director of AI at Arm, who offered the students a grounded, strategic view of the shifting landscape, the commoditization of coding, and the bifurcation of the AI industry.
Here are my notes from the video. They’re a good guide, but the video is so packed with info that you really should watch it to get the most from it!
The golden age of the “product engineer”
Ng opened the session with optimism, saying that this current moment is the “best time ever” to build with AI. He cited research suggesting that every 7 months, the complexity of tasks AI can handle doubles. He also argued that the barrier to entry for building powerful software has collapsed.
Speed is the new currency! The velocity at which software can be written has changed largely due to AI coding assistants. Ng admitted that keeping up with these tools is exhausting (his “favorite tool” changes every three to six months), but it’s non-negotiable. He noted that being even “half a generation behind” on these tools results in a significant productivity drop. The modern AI developer needs to be hyper-adaptive, constantly relearning their workflow to maintain speed.
The bottleneck has shifted to what to build. As writing code becomes cheaper and faster, the bottleneck in software development shifts from implementation to specification.
Ng highlighted a rising trend in Silicon Valley: the collapse of the Engineer and Product Manager (PM) roles. Traditionally, companies operated with a ratio of one PM to every 4–8 engineers. Now, Ng sees teams trending toward 1:1 or even collapsing the roles entirely. Engineers who can talk to users, empathize with their needs, and decide what to build are becoming the most valuable assets in the industry. The ability to write code is no longer enough; you must also possess the product instinct to direct that code toward solving real problems.
The company you keep: Ng’s final piece of advice focused on network effects. He argued that your rate of learning is predicted heavily by the five people you interact with most. He warned against the allure of “hot logos” and joining a “company of the moment” just for the brand name and prestige-by-association. He shared a cautionary tale of a top student who joined a “hot AI brand” only to be assigned to a backend Java payment processing team for a year. Instead, Ng advised optimizing for the team rather than the company. A smaller, less famous company with a brilliant, supportive team will often accelerate your career faster than being a cog in a prestigious machine.
Surviving the market correction
Ng handed over the stage to Moroney, who started by presenting the harsh realities of the job market. He characterized the current era (2024–2025) as “The Great Adjustment,” following the over-hiring frenzy of the post-pandemic boom.
The three pillars of success To survive in a market where “entry-level positions feel scarce,” Moroney outlined three non-negotiable pillars for candidates:
- Understanding in depth: You can’t just rely on high-level APIs. You need academic depth combined with a “finger on the pulse” of what is actually working in the industry versus what is hype.
- Business focus: This is the most critical shift. The era of “coolness for coolness’ sake” is over. Companies are ruthlessly focused on the bottom line.Moroney put a spin on the classic advice, “Dress for the job you want, not the job you have,” and suggested that if you’re a job-seeker, that you “not let your output be for the job you have, but for the job you want.” He based this on his own experience of landing a role at Google not by preparing to answer brain teasers, but by building a stock prediction app on their cloud infrastructure before the interview.
- Bias towards delivery: Ideas are cheap; execution is everything. In a world of “vibe coding” (a term he doesn’t like — he prefers something more line “prompting code into existence” or “prompt coding”), what will set you apart is the ability to actually ship reliable, production-grade software.
The trap of “vibe coding” and technical debt: Moroney addressed the phenomenon of using LLMs to generate entire applications. They may be powerful, but he warned that they also create massive “technical debt.”
The 4 Realities of Modern AI Work
Moroney outlined four harsh realities that define the current workspace, warning that the “coolness for coolness’ sake” era is over. These realities represent a shift in what companies now demand from engineers.
Business focus is non-negotiable. Moroney noted a significant cultural “pendulum swing” in Silicon Valley. For years, companies over-indexed on allowing employees to bring their “whole selves” to work, which often prioritized internal activism over business goals. That era is ending. Today, the focus is strictly on the bottom line. He warned that while supporting causes is important, in the professional sphere, “business focus has become non-negotiable.” Engineers must align their output directly with business value to survive.
2. Risk mitigation is the job. When interviewing, the number one skill to demonstrate is not just coding, but the ability to identify and manage the risks of deploying AI. Moroney described the transition from heuristic computing (traditional code) to intelligent computing (AI) as inherently risky. Companies are looking for “Trusted Advisors” who can articulate the dangers of a model (hallucinations, security flaws, or brand damage) and offer concrete strategies to mitigate them.
3. Responsibility is evolving. “Responsible AI” has moved from abstract social ideals to hardline brand protection. Moroney shared a candid behind-the-scenes look at the Google Gemini image generation controversy (where the model refused to generate images of Caucasian people due to over-tuned safety filters). He argued that responsibility is no longer just about “fairness” in a fluffy sense; it is about preventing catastrophic reputational damage. A “responsible” engineer now ensures the model doesn’t just avoid bias, but actually works as intended without embarrassing the company.
4. Learning from mistakes is constant. Because the industry is moving so fast, mistakes are inevitable. Moroney emphasized that the ability to “learn from mistakes” and, crucially, to “give grace” to colleagues when they fail is a requirement. In an environment where even the biggest tech giants stumble publicly (as seen with the Gemini launch), the ability to iterate quickly after a failure is more valuable than trying to be perfect on the first try.
Technical debt
Just like a mortgage, debt isn’t inherently bad, but you must be able to service it. He defined the new role of the senior engineer as a “trusted advisor.” If a VP prompts an app into existence over a weekend, it is the senior engineer’s job to understand the security risks, maintainability, and hidden bugs within that spaghetti code. You must be the one who understands the implications of the generated code, not just the one who generated it.
The dot-com parallel: Moroney drew a sharp parallel between the current AI frenzy and the Dot-Com bubble of the late 1990s. He acknowledged that while we are undoubtedly in a financial bubble, with venture capital pouring billions into startups with zero revenue, he emphasizes that this does not imply the technology itself is a sham.
Just as the internet fundamentally changed the world despite the 2000 crash wiping out “tourist” companies, AI is a genuine technological shift that is here to stay. He warns students to distinguish between the valuation bubble (which will burst) and the utility curve (which will keep rising), advising them to ignore the stock prices and focus entirely on the tangible value the technology provides.
The bursting of this bubble, which Moroney terms “The Great Adjustment,” marks the end of the “growth at all costs” era. He argues that the days of raising millions on a “cool demo” or “vibes” are over. The market is violently correcting toward unit economics, meaning AI companies must now prove they can make more money than they burn on compute costs. For engineers, this signals a critical shift in career strategy: job security no longer comes from working on the flashiest new model, but from building unglamorous, profitable applications that survive the coming purge of unprofitable startups.
Future-proofing: “Big AI ” vs. “Small AI”
Perhaps the most strategic insight from the lecture was Moroney’s prediction of a coming “bifurcation” in the AI industry over the next five years.
The industry is splitting into two distinct paths:
- “Big AI”: The AI made by massive, centralized players such as OpenAI, Google, and Anthropic, who are chasing after AGI. This relies on ever-larger models hosted in the cloud.
- “Small AI”: AI systems that are based on open-weight (he prefers “open-weight” to “open source” when describing AI models), self-hosted, and on-device models. Moroney also calls this “self-hosted AI.”
Moroney is bullish on “Small AI.” He explained that many industries are very protective of their intellectual property, such as movie/television studios and law firms. These business will will never send their intellectual property to a centralized model like GPT-4 due to privacy and IP concerns. This creates a massive, underserved market for engineers who can fine-tune small models to run locally on a device or private server.
Moroney urged the class to diversify their skills. Don’t just learn how to call an API; learn how to optimize a 7-billion parameter model to run on a laptop CPU. That is where the uncrowded opportunity lies.
Agentic Workflows: The “How” of Future Engineering: Moroney’s advice was to stop thinking of agents as magic and start treating them as a rigorous engineering workflow consisting of four steps:
- Intent: Understanding exactly what the user wants.
- Planning: Breaking that intent down into steps.
- Tools: Giving the model access to specific capabilities (search, code execution).
- Reflection: Checking if the result met the intent. He shared a demo of a movie-making tool where simply adding this agentic loop transformed a hallucinated, glitchy video into a coherent scene with emotional depth.
Conclusion: Work hard
I’ll conclude this set of notes with what Ng said at the conclusion of his introduction to the lecture, which he described as “politically incorrect”: Work hard.
While he acknowledged that not everyone is in a situation where they can do so, he pointed out that among his most successful PhD students, the common denominator was an incredible work ethic: nights, weekends, and the “2 AM hyperparameter tuning.”
In a world drowning in hype, Ng’s and Moroney’s “brutally honest” playbook is actually quite simple:
- Use the best tools to move fast
- Understand the business problem you’re trying to solve, and understand it deeply.
- Ignore the noise of social media and the trends being hyped there. Build things that actually work.
- And finally, to quote Ng: “Between watching some dumb TV show versus finding your agentic coder on a weekend to try something… I’m going to choose the latter almost every time.”