Get Clips From Video > 60 seconds

The Death of Software 1.0

Clip Title: Why Software 1.0 is Dying

Primary Source Timecode Range: [00:02:52] - [00:03:34]

Script Body:

Hook: "Software 1.0, I'm writing code. Software 3.0... your programming now turns to prompting."

• [00:02:52] "So software 1.0, I'm writing code, software 2.0, I'm actually programming by creating data sets and training neural networks."
• [00:03:04] "And then what happened is that basically if you train one of these GPT models or LLMs on a sufficiently large set of tasks... these actually become kind of like a programmable computer in a certain sense."
• [00:03:20] "So software 3.0 is kind of about, you know, your programming now turns to prompting and what's in the context window is your lever over the interpreter that is the LLM."
• [00:03:31] "That is kind of like interpreting your context and performing computation in the digital information space."

Total clip runtime: ~42 seconds

Bonus Quotes on This Topic:

• "Software 1.0 was explicit rules, software 2.0 was learned weights, software 3.0 is this." [00:02:38] (Perfect alt hook/intro).
• "What is the piece of text to copy paste to your agent? That's the programming paradigm now." [00:04:46] (Great for a punchy summary line).
• "The agent has its own intelligence that it packages up and then it kind of like follows the instructions." [00:04:29] (Explains the 'why' for B-roll).

The 'Behind' Feeling

Clip Title: Why Andrej Karpathy Feels 'Behind'

Primary Source Timecode Range: [00:01:05] - [00:01:42]

Script Body:

Hook: "I've never felt more behind as a programmer. I just started to notice that with the latest models... the chunks just came out fine."

• [00:01:05] "I've been using agentic tools like lot code, adjacent things for a while... and it was very good at you know chunks of code and sometimes it would mess up."
• [00:01:17] "I would say December was this clear point where... I just started to notice that with the latest models the chunks just came out fine and then I kept asking for more and it just came out fine."
• [00:01:31] "And then I can't remember the last time I corrected it and then I was, I just, you know, trusted the system more and more and then I was vibe coding."
• [00:01:39] "So it was kind of a, I do think that it was a very stark transition."

Total clip runtime: ~38 seconds

Bonus Quotes on This Topic:

• "Things have changed fundamentally and especially on this like agentic coherent workflow that really started to actually work." [00:01:57] (Reinforces the 'December' shift).
• "My side projects folder is like extremely full with lots of random things and, uh, just, uh, vibe coding all the time." [00:02:16] (Shows the result of the shift).
• "You really had to look again and you had to look as of December because things have changed fundamentally." [00:01:54] (Strong call to action for viewers).

The Car Wash Paradox

Clip Title: The Jagged Intelligence Paradox

Primary Source Timecode Range: [00:11:15] - [00:11:53]

Script Body:

Hook: "How is it possible that state-of-the-art models will refactor a 100,000 line codebase... and yet tell me to walk to a car wash?"

• [00:11:15] "I think to me the big mystery is... the favorite example for a while was how many letters are in a strawberry and the models would famously get this wrong."
• [00:11:28] "But the new one is: I want to go to a car wash to wash my car and it's 50 meters away. Should I drive or should I walk?"
• [00:11:36] "State-of-the-art models today will tell you to walk because it's so close. This is insane."
• [00:11:42] "How is it possible that Opus 4.7 will simultaneously refactor a 100,000 line codebase or find zero day vulnerabilities and yet tells me to walk to this car wash?"

Total clip runtime: ~40 seconds

Bonus Quotes on This Topic:

• "To whatever extent these models remain jagged, it's an indication that number one maybe something's slightly off." [00:11:54] (Insightful follow-up).
• "They end up basically creating these like jagged entities that really peak in capability in kind of like verifiable domains." [00:10:31] (Technical explanation for B-roll).
• "They are these like spiky entities. They're a bit fallible, a little bit stochastic, but they are extremely powerful." [00:16:31] (Great characterization of AI).

Agentic Engineering vs. Vibe Coding

Clip Title: Vibe Coding vs. Agentic Engineering

Primary Source Timecode Range: [00:15:58] - [00:16:42]

Script Body:

Hook: "Vibe coding is about raising the floor for everyone... Agentic engineering is about preserving the quality bar."

• [00:15:58] "I would say vibe coding is about raising the floor for everyone in terms of what they can do in software. Everyone can vibe code anything and that's amazing."
• [00:16:07] "But then I would say agentic engineering is about preserving the quality bar of what existed before in professional software."
• [00:16:13] "You're not allowed to introduce vulnerabilities due to vibe coding. You're still responsible for your software just as before, but can you go faster?"
• [00:16:26] "When I call it that because I do think it's kind of like an engineering discipline. How do you coordinate them to go faster without sacrificing your quality bar?"

Total clip runtime: ~44 seconds

Bonus Quotes on This Topic:

• "People used to talk about the 10x engineer previously I think that this is magnified a lot more. 10x is not the speed up you gain." [00:17:01] (Provocative take on productivity).
• "People who are very good at this peak a lot more than 10x from my perspective right now." [00:17:11] (Strong ending/payoff quote).
• "One is about maybe raising the floor and the other is about extrapolating." [00:16:48] (Summarizes the contrast).

Thinking vs. Understanding

Clip Title: You Can't Outsource Understanding

Primary Source Timecode Range: [00:28:05] - [00:28:46]

Script Body:

Hook: "You can outsource your thinking, but you can't outsource your understanding."

• [00:28:05] "There was a tweet that blew my mind recently... You can outsource your thinking but you can't outsource your understanding. I think that's really nicely put."
• [00:28:21] "Because I'm still part of the system and I still have to somehow, information still has to make it into my brain."
• [00:28:27] "I feel like I'm becoming a bottleneck of just even knowing what are we trying to build, why is it worth doing, how do I direct my agents."
• [00:28:34] "I do still think that ultimately something has to direct the thinking and the processing and that's still kind of fundamentally constrained somehow by understanding."

Total clip runtime: ~41 seconds

Bonus Quotes on This Topic:

• "The LLMs certainly don't excel at understanding. You still are uniquely in charge of that." [00:29:22] (Clarifies the human role).
• "What still remains worth learning deeply when intelligence gets cheap?" [00:27:58] (Great setup question/alt hook).
• "You can't be a good director if you don't understand, because the LLMs certainly don't excel at it." [00:29:20] (Reinforces the leadership analogy).

Coding is Solved

50% Solved, 100% Possible

Boris explains why he believes the era of manual coding is effectively over and how language choice no longer limits the model's capabilities.

Script Body:

• Hook: "You’ve said on the record that you think coding is solved." [00:05:09.040]
• Line 2: "I mean, for me, it's 100%. Like, the model writes 100% of my code." [00:05:41.160]
• Line 3: "The model was not as intelligent as it is today, so the language and the framework mattered a lot." [00:05:59.480]
• Line 4: "Nowadays, you know, it can write whatever, and it can pick up new languages, new frameworks it hasn't seen." [00:06:03.120]
• Line 5: "Usually the answer is just wait for the next model." [00:06:47.440]

Total Runtime: ~42 seconds

Bonus Quotes on This Topic:

• "Who writes 100% of their code using a agent like Claude code? Okay. So, like 50% solved." [00:05:27.520] — Great for an interactive/funny alt hook.
• "There was a day last week I did like 150 PRs in a day. That was a record." [00:06:28.560] — Aha! insight showing the scale of modern engineering.
• "For us, this happened sometime in October, November last year." [00:06:18.520] — Visual timeline context.

The Slash Loop Future

The Future is Loops

Boris reveals his secret personal setup and why "Loops" are the simplest, most powerful way to command agents to do background work.

Script Body:

• Hook: "I sort of feel like loops are the future at this point." [00:08:35.679]
• Line 2: "It's like the coolest thing. It's like the simplest thing that works." [00:07:59.840]
• Line 3: "All it is is you have Claude use cron to schedule a job for some point in the future, and it's a repeat job." [00:08:02.760]
• Line 4: "I have one that's babysitting my PRs, like fixing CI, auto-rebasing. I have another one that grabs feedback from Twitter and clusters it for me every 30 minutes." [00:08:18.120]
• Line 5: "If you haven't experimented with it, highly highly recommend it." [00:08:39.520]

Total Runtime: ~38 seconds

Bonus Quotes on This Topic:

• "Currently probably like a few hundred agents going. Usually every night I have like a few thousand that are doing kind of deeper work." [00:07:40.640] — B-roll voiceover for scale.
• "The thing that I've been finding myself using more and more is the loop. So, this is {slash} loop." [00:07:52.520] — Instructional/Practical tip.
• "It can run every every minute, every 5 minutes, every day, kind of however often you want to schedule it." [00:08:09.480] — Clarifies nuance.

The Rise of the Polymath

Everyone is an Engineer Now

A look into the "Claude Code" team structure where finance guys, designers, and researchers are all writing production-level code.

Script Body:

• Hook: "There's going to be a lot more generalists than there are today." [00:09:15.280]
• Line 2: "The thing that we're going to start to see a lot more of is generalists that are cross-disciplinary." [00:09:35.280]
• Line 3: "This is engineers that are really good at product engineering, but also really great at design. Or really great at product and data science." [00:09:40.720]
• Line 4: "Everyone on our team codes. Our engineering manager, our product manager, our designers, our finance guy... every single person on our team writes code." [00:09:58.440]
• Line 5: "They're specialist in something, but now also everyone's just coding." [00:10:10.800]

Total Runtime: ~45 seconds

Bonus Quotes on This Topic:

• "I think building software becoming a skill like... 'I know how to send a text message.'" [00:15:33.400] — Strong alt hook.
• "You don't need a degree in reading and writing to know how to read and write." [00:16:43.079] — Revelatory analogy.
• "The best person to write accounting software involves knowing the domain really well and coding is the easy part." [00:17:05.160] — Supportive context.

The Coding Printing Press

Software's Gutenberg Moment

Boris draws a powerful historical parallel between the invention of the printing press and the democratization of software creation.

Script Body:

• Hook: "Software will be a thing that is fully democratized, that anyone can do." [00:16:55.440]
• Line 2: "In the 1400s, the printing press in Europe. Before the printing press, essentially 10% of the European population was literate." [00:15:52.040]
• Line 3: "The printing press was invented... in the 50 years after the first printing press, there was more literature published in Europe than in the thousand years before." [00:16:14.720]
• Line 4: "The cost of a book went down like a 100x." [00:16:25.120]
• Line 5: "The thing that's about to happen and it's going to be much faster than 50 years is software will be a thing that is fully democratized." [00:16:51.440]

Total Runtime: ~43 seconds

Bonus Quotes on This Topic:

• "Learning to read and write is hard. You need education systems... But over the next few hundred years, literacy globally went up to like 70%." [00:16:30.280] — Deep narrative voiceover.
• "It's knowing the domain that's the hard part. And I think this is just obviously the future." [00:17:15.560] — Final punchy closer.

The Startup Advantage

The Best Time to Build

Why small startups can now go head-to-head with giants, and how internal resistance in big companies is your biggest competitive advantage.

Script Body:

• Hook: "I don't know, it's the best time to build. It's the best time to be a startup." [00:12:48.960]
• Line 2: "The number of startups in the next 10 years that are just going to like disrupt everything is going to increase like 10x." [00:12:13.440]
• Line 3: "You could build a thing that's as valuable as a large company and you can actually compete head-to-head." [00:12:25.320]
• Line 4: "The large company has to evolve their business process, they have to retrain everyone... they're going to face a lot of internal resistance to that." [00:12:31.880]
• Line 5: "If you’re starting fresh, then you can kind of build with AI natively from the ground up." [00:12:44.040]

Total Runtime: ~40 seconds

Bonus Quotes on This Topic:

• "There's so much disruption coming." [00:12:53.280] — Final sting.
• "We have no more manually written code anywhere at the company. All of the SQL is written by models." [00:18:30.880] — Evidence-based hook.
• "Some of these modes are going to get more important and some are going to get less important." [00:11:28.120] — Introductory context.

The 'Zip File' Mental Model

Title: LLMs Are Simple Zip Files

Primary Source Timecode Range: [00:12:15 - 00:12:48] Total Clip Runtime: 42 Seconds

Script:

Hook: [00:12:22] "Hi, I'm ChatGPT. I am a one terabyte zip file."

[00:12:26 - 00:12:35] "My knowledge comes from the internet which I read in its entirety about six months ago and I only remember vaguely, okay?"

[00:12:37 - 00:12:43] "And my winning personality was programmed by example by human labelers at OpenAI. So the personality is programmed in post-training."

[00:12:44 - 00:12:48] "And the knowledge comes from compressing the internet during pre-training. And this knowledge is a little bit out of date."

[00:08:24 - 00:08:31] "The zip file is not exact. The zip file is a lossy and probabilistic zip file because we can't possibly represent all of internet in just one terabyte."

[00:11:58 - 00:12:10] "You're talking to a zip file. If you stream tokens to it, it will respond with tokens back. It has the knowledge from pre-training and it has the style and form from post-training."

Bonus Quotes on This Topic:

• Quote: "Think of it as a one terbyte file on a disk secretly that represents one trillion parameters." [01:11:33] (Clarifies technical scale)
• Quote: "What you are talking to is a fully self-contained entity by default." [01:11:28] (Strong alt hook)
• Quote: "It's trying to basically take tokens and it's trying to predict the next token in a sequence... it's kind of like this internet document generator." [00:09:03] (Explains mechanism)

The Context Window Limit

Title: Stop Being Lazy with Your Chat History

Primary Source Timecode Range: [01:16:32 - 01:17:53] Total Clip Runtime: 45 Seconds

Script:

Hook: [01:16:34] "Anytime you are switching topic, I encourage you to always start a new chat."

[01:16:39 - 01:16:44] "When you start a new chat, you are wiping the context window of tokens and resetting it back to zero again."

[01:16:50 - 01:16:55] "I encourage you to do this because these tokens in this window are expensive. Number one, the model can actually find it a little bit distracting."

[01:17:15 - 01:17:18] "It could actually decrease the accuracy of the model and of its performance."

[01:17:38 - 01:17:46] "Think of the tokens in the context window as a precious resource. Think of that as the working memory of the model and don't overload it with irrelevant information."

[01:06:50 - 01:07:00] "When I click new chat here, that wipes the token window. That resets the tokens to basically zero again and restarts the conversation from scratch."

Bonus Quotes on This Topic:

• Quote: "Everything together we are building out a token window... which we also call the context window." [01:06:34] (Great for B-roll voiceover)
• Quote: "Your model is actually slightly slowing down. It's becoming more expensive to calculate the next token." [01:17:28] (Reasoning for performance)
• Quote: "Anything that is inside this context window is kind of like in the working memory of this conversation." [01:07:44] (Simplifies the concept)

The 'Thinking' Model Revolution

Title: Why You Need 'Thinking' Models

Primary Source Timecode Range: [00:23:42 - 00:25:28] Total Clip Runtime: 44 Seconds

Script:

Hook: [00:22:58] "The next topic I want to now turn to is that of 'thinking models' quote unquote."

[00:23:42 - 00:23:51] "The model will try out different ideas, it will backtrack, it will revisit assumptions. It's only in the reinforcement learning that the model can find the thinking process that works for it."

[00:24:50 - 00:24:58] "Qualitatively, the model will do a lot more thinking and what you can expect is that you will get higher accuracies especially on problems that are math and code."

[00:25:12 - 00:25:21] "The models will do thinking and that can sometimes take multiple minutes because the models will emit tons and tons of tokens. You have to wait because the model is thinking."

[00:25:23 - 00:25:28] "In situations where you have very difficult problems, this might translate to higher accuracy."

[00:29:43 - 00:29:51] "These models are most effective for difficult problems in math and code and things like that. In those kinds of cases, they can push up the accuracy of your performance."

Bonus Quotes on This Topic:

• Quote: "These thinking strategies... very much resemble kind of the inner monologue you have when you go through problem solving." [00:23:30] (Revelatory insight)
• Quote: "Try out the non-thinking models because their responses are really fast, but when I suspect the response is not as good... I will change it to a thinking model." [00:30:10] (Practical tip)
• Quote: "Wait a few minutes and then also comes up with the correct answer." [00:29:33] (Sets user expectations)

Don't Read Books Alone

Title: Stop Reading Books Alone

Primary Source Timecode Range: [00:55:01 - 00:55:31] Total Clip Runtime: 38 Seconds

Script:

Hook: [00:55:03] "It is rarely ever the case anymore that I read books just by myself. I always involve an LLM to help me read."

[00:55:31 - 00:55:37] "You basically pull up the book and you have to get access to the raw content. You can remind the model by loading this into the context window."

[00:57:39 - 00:57:46] "I find that basically going hand in hand with LLMs dramatically increases my retention and my understanding of these chapters."

[00:58:11 - 00:58:16] "LLMs make a lot of reading very dramatically more accessible than it used to be before because you’re not just right away confused."

[00:58:18 - 01:00:00] "You can actually kind of go slowly through it and figure it out together with the LLM in hand. I encourage you to experiment with it and... don't read books alone."

Bonus Quotes on This Topic:

• Quote: "I would feel a lot more courage approaching a very old text that is outside of my area of expertise." [00:58:02] (Emotional benefit)
• Quote: "You can upload these documents to the LLM... and basically read the document together." [00:52:35] (Clarifies workflow)
• Quote: "I'm not aware of a tool that makes this very easy... I do this clunky back and forth." [00:58:43] (Adds authenticity/vulnerability)

The 'Council of LLMs' Strategy

Title: Create Your Own 'LLM Council'

Primary Source Timecode Range: [00:22:36 - 00:22:54] Total Clip Runtime: 40 Seconds

Script:

Hook: [00:22:42] "I kind of refer to all these models as my LLM Council."

[00:22:36 - 00:22:42] "I end up personally just paying for a lot of them and then asking all of them the same question."

[00:21:57 - 00:22:03] "I like to go between different models and asking them similar questions and seeing what they think about."

[00:22:45- 00:22:51] "They’re kind of like the Council of language models. If I'm trying to figure out where to go on a vacation, I will ask all of them."

[00:01:21 - 00:01:31] "In some cases there are kind of like unique experiences that are not found in ChatGPT. We’re going to see examples of those."

[01:04:17 - 01:04:26] "The state of the LLMs right now is such that different LLMs have different tools available to them and you kind of have to keep track of it."

Bonus Quotes on This Topic:

• Quote: "Experiment with different providers, experiment with different pricing tiers for the problems that you are working on." [01:22:28] (Strong alt hook)
• Quote: "One of the most feature-rich... but there are many other kind of clones available I would say." [00:01:16] (Contextualizes the market)
• Quote: "Find the one that works best for you." [01:22:25] (Actionable closing line)