7 INSANE loops you need to try right now
Credibility score: 49/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "7 INSANE loops you need to try right now" and rated it 49/100 for credibility (a BS score of 51/100 — mixed credibility), on 2026-06-23. Its weakest claim — "Anticipates and dismisses a common objection with a rhetorical question." — scored 20/100 and was flagged as straw man. 19 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 19 claims analyzed: 1 scored under 40, 15 between 40 and 69, and 3 at 70 or above.
Claims analyzed
Declares 'loops' the 'single biggest unlock' for AI software development. — Loaded Language (45/100)
At 0:00
Calling something the 'single biggest unlock' is pure hype, not a technical spec. It's like saying 'this is the most delicious water ever!' 💧
Why this score: The phrase 'single biggest unlock' uses highly enthusiastic, superlative language to create excitement and a sense of urgency around 'loops.' It's an emotional appeal rather than a precise, measurable claim about their actual impact or superiority over other methods. It sets a very high expectation without providing any comparative data or specific metrics to back it up.
Original quote: “Loops are emerging as the single biggest unlock for people building software with artificial intelligence right now.”
Claims 'most people don't even know what loops are' in AI. — Confidence Mismatch (45/100)
At 0:05
Says 'most people' don't know, but how many people did he poll? Just vibes and a broad generalization. 🤷♂️
Why this score: The speaker makes a confident assertion about the general knowledge of 'most people' regarding AI loops without providing any data, survey results, or specific evidence to support this claim. It's a broad generalization that assumes a lack of understanding, which might be true for a general audience but is presented as a definitive fact without substantiation.
Original quote: “But most people don't even know what loops are.”
Defines a loop as an AI agent working autonomously towards a specified goal. — No Frame (75/100)
At 0:21
A pretty standard, if slightly simplified, definition of a loop in an AI context. It's a foundational concept. 🤓
Why this score: The definition provided is consistent with how 'loops' are generally understood in programming and AI, referring to iterative processes or autonomous agents working towards a goal. It's a factual statement defining a concept, without any apparent framing tricks.
Original quote: “A loop is a way to allow your AI coding agent to work autonomously towards a specified”
Loops remove humans for faster agent work, presented as a key benefit. — Loaded Language (45/100)
At 0:31
Removing humans sounds great for efficiency, but it also sounds like a job-killer. The framing is all upside, no downside 🤖💸
Why this score: The speaker uses 'removes humans' as a positive, implying efficiency, but it's loaded language that glosses over potential downsides like job displacement or loss of human oversight. It's framed purely as an advantage without acknowledging the full implications.
Original quote: “The most important thing about loops is that it removes humans that allows the agent to work much more quickly towards this defined goal.”
Simplifying loops to just 'trigger and goal' for completion. — No Frame (75/100)
At 0:46
Breaking down the core components of a loop into just a trigger and a goal. Straightforward enough. ✅
Why this score: This is a clear and concise explanation of the fundamental components of a loop, which are indeed a trigger to start it and a goal to define its completion. No obvious framing or manipulation here; it's a basic technical definition.
Original quote: “Well, you need
two things. You need a trigger and you
need a goal. With those two things, you
can complete the loop.”
Defining a loop requires only a trigger and a goal. — No Frame (75/100)
At 0:46
Just laying out the basic components of a loop. No tricks here, just the setup. ⚙️
Why this score: This is a straightforward explanation of the fundamental components required for a loop in this context. There's no apparent framing or rhetorical device being used beyond a clear definition.
Original quote: “Well, you need two things. You need a trigger and you need a goal. With those two things, you can complete the loop.”
Goals are either 'verifiable' or 'LLM as a judge,' giving the model self-determination. — Confidence Mismatch (45/100)
At 1:28
Giving an LLM 'the ability to determine when it has reached the goal' sounds like a recipe for a self-congratulatory AI. What could go wrong? 😂
Why this score: The speaker confidently presents 'LLM as a judge' as a valid goal type, which essentially means the AI decides when it's done. This is a significant leap of faith in AI autonomy, and the confidence in this approach seems mismatched with the potential for the AI to declare success prematurely or incorrectly without external verification. It's a bit like letting the fox guard the hen house and then asking the fox if it's done guarding.
Original quote: “And for the goal, the goal can be basically one of two things. It can be verifiable or we can use LLM as a judge. So if it's verifiable, it is something concrete, some specific number or some way to test it deterministically. If it is LLM as a judge, that means we're giving the model the ability to…”
LLM as a judge means the AI decides when 'refactored enough' is achieved. — Just Vibes (50/100)
At 2:07
So the LLM just gets to decide when it's 'satisfied'? That's not a metric, that's a mood. 💅🤖
Why this score: This is a prime example of the 'LLM as a judge' concept, where 'satisfaction' is entirely subjective and determined by the AI itself. It's less a concrete claim and more a demonstration of a potentially vague and self-serving goal for an AI, which is ripe for comedic observation rather than a strict framing analysis.
Original quote: “And for LLM as a judge, one example would be refactor until satisfied. And the satisfaction just means you as the LLM get to determine when we are satisfactorily refactored enough.”
Using 'refactor until satisfied' as an example of an LLM-judged goal, implying subjective completion. — Confidence Mismatch (45/100)
At 2:07
LLM deciding 'satisfactorily refactored enough'? That's a lot of trust in a machine's aesthetic judgment. What could go wrong? 😬🔥
Why this score: While 'LLM as a judge' is a valid concept, using 'refactor until satisfied' as a prime example highlights a significant confidence mismatch. 'Satisfaction' and 'enough' are highly subjective human concepts. Giving an LLM the sole authority to determine 'satisfactory refactoring' without human oversight or objective metrics is a bold claim that glosses over the potential for the AI to go off-track or produce suboptimal results, as hinted by community comments about AI getting 'obsessed' or 'wasting time going nowhere.'
Original quote: “And for LLM as a judge,
one example would be refactor until
satisfied. And the satisfaction just
means you as the LLM get to determine
when we are satisfactorily
refactored enough.”
Launching a "free" loop library — presented as a helpful tool, but it's a lead magnet. — Plain Sales Pitch (45/100)
At 2:30
It's "free" but the whole point is to get you to click the link and engage. Classic lead gen move. 🎣
Why this score: The speaker is presenting a 'free' resource, which is a common tactic to build an audience or generate leads for future paid offerings. While genuinely free, it's still a promotional tool for his brand/platform.
Original quote: “actually give concrete use cases. And I wanted to fix this. That is why I am launching the loop library. It is a free library. I'm basically taking all the loops that I use and the ones that I see other people use and putting them in a single place so you can see them. You can be inspired by them…”
AI will 'continuously optimize' until page loads are under 50ms — a confidence mismatch. — Confidence Mismatch (45/100)
At 2:30
Claiming AI will 'continuously optimize' until a specific, perfect outcome is achieved is a bold promise. AI isn't magic, it's a tool. ✨
Why this score: The speaker expresses high confidence in the AI's ability to 'continuously optimize' until a very specific, challenging goal (sub-50ms page loads across an entire app) is met. While AI can assist, the absolute certainty that it 'will' achieve this without human intervention or potential roadblocks is a strong claim that often overestimates current AI capabilities in complex, open-ended optimization tasks. It's a bit like saying a self-driving car will 'continuously drive' until it wins the Indy 500.
Original quote: “load it. If it's above 50 milliseconds, it's going to continuously optimize it until it gets it under 50 milliseconds.”
AI optimized pages to load under 50ms in production. — Confidence Mismatch (45/100)
At 4:30
Claiming "every page" was optimized to under 50ms in production is a huge leap without showing any actual data. That's a big "trust me, bro" moment. 🤡
Why this score: The speaker states that 'every page' was optimized to load under 50 milliseconds on production, but provides no actual evidence, metrics, or a demo of this achievement. It's a bold claim without the receipts.
Original quote: “I sent this as a goal. Look for more optimizations to make sure every page loads in under 50 milliseconds on production.”
LLM as a judge for documentation coverage, admitting no verifiable way. — Confidence Mismatch (45/100)
At 6:30
Says there's 'no verifiable way' to know if docs are complete, then immediately pivots to 'LLM, you decide.' So, we're just trusting the vibes of an AI? 🤖🤷♂️
Why this score: The speaker acknowledges a fundamental lack of verifiable metrics for complete documentation coverage, yet proposes using an LLM as a 'judge' to decide. This creates a confidence mismatch, as the solution relies on an unverified authority to solve an unverifiable problem.
Original quote: “This is an example of LLM as a judge. There's no verifiable way to know if we have complete documentation coverage. There may be some ways that we can say, okay, as long as a piece of documentation covers this section of the code, but ultimately what we're doing is saying, okay, LLM, you decide.”
Using LLMs to 'judge' documentation completeness because there's 'no verifiable way' otherwise. — Confidence Mismatch (45/100)
At 6:30
Says there's 'no verifiable way' to check docs, then immediately pivots to 'LLM, you decide.' That's not verification, that's outsourcing the problem! 🤖🤷♂️
Why this score: The speaker claims there's 'no verifiable way' to assess documentation completeness, then proposes using an LLM to 'decide.' This implies the LLM provides a verifiable solution, which contradicts the initial statement. It's a leap of faith, not a verifiable method.
Original quote: “This is an example of LLM as a judge. There's no verifiable way to know if we have complete documentation coverage. There may be some ways that we can say, okay, as long as a piece of documentation covers this section of the code, but ultimately what we're doing is saying, okay, LLM, you decide.”
Defining 'happy with the architecture' as a goal for an LLM loop, using a subjective metric. — Confidence Mismatch (45/100)
At 8:30
Setting 'happy with the architecture' as a goal for an AI is like asking your dog to write a symphony. Good luck defining 'happy' for a machine! 🤖🤷♀️
Why this score: The speaker presents 'happy with the architecture' as a clear goal for an LLM, but 'happy' is inherently subjective and non-quantifiable for an AI. While they later mention giving more guidance, the initial framing implies a level of autonomous understanding the LLM doesn't possess, creating a confidence mismatch between the task and the tool's capability.
Original quote: “architecture. Here is the trigger and the goal all in one sentence. Refactor, which is what the loop is going to do, until you are happy with the architecture. Happy with the architecture is the goal.”
Declaring 'no more unaddressed errors' as a concrete goal for the loop. — Confidence Mismatch (45/100)
At 10:30
Saying 'no more unaddressed errors' is a goal, not a guarantee. That's a bold promise for a 'loop' to achieve. 🤖✨
Why this score: The speaker presents 'no more unaddressed errors' as a 'very concrete goal' for the loop. While it's an aspirational goal, stating it as a concrete outcome for an automated process oversimplifies the complexities of software development and error resolution, implying a level of certainty that's hard to achieve in practice.
Original quote: “no more unressed errors in the logs. So that is a very concrete goal for this loop.”
Anticipates and dismisses a common objection with a rhetorical question. — Straw Man (20/100)
At 12:30
Sets up a straw man by saying 'you might be thinking it's just tests' — then immediately pivots to why it's 'not just tests.' Classic move. 🤡
Why this score: The speaker preemptively addresses a potential counter-argument ('just sounds like tests') only to immediately differentiate their concept. This is a form of straw man, where a simplified or slightly mischaracterized version of an opposing view is presented and then easily refuted, making the speaker's actual point seem stronger by comparison.
Original quote: “Now, a lot of you might be thinking, "Wow, that just sounds like tests, right? It's just like a test suite.”
Claiming 'non-deterministic' LLM loops 'really do work' for optimization. — Confidence Mismatch (45/100)
At 12:30
Says it 'really does work' but the 'non-deterministic' part means results can vary. That's a lot of confidence for something that's literally unpredictable. 🎲
Why this score: The speaker uses strong, definitive language ('really does work') to describe a system that they explicitly state is 'non-deterministic'. Non-deterministic systems, by definition, don't always produce the same output for the same input, making a blanket statement of 'it really does work' a mismatch with the inherent variability. It's a leap of faith, not a guarantee.
Original quote: “Well, kind of. But this is actually non-deterministic. This is allowing the model to go through every single use case in your application, in your product, figure out if it's good enough, determined by the LLM, and update it if necessary. This one really does work.”
Speaker admits not knowing how AI will decide features, yet still uses loops. — Confidence Mismatch (45/100)
At 14:30
Says he doesn't know what the AI will do, but still recommends loops. That's a 'trust me, bro' with extra steps 😬
Why this score: The speaker acknowledges a significant unknown (AI's decision-making on features) but proceeds to discuss using loops, creating a mismatch between the expressed uncertainty and the implied confidence in the utility of loops for feature building. It's like saying 'I don't know where this car is going, but get in!'
Original quote: “I have not really found a way to build features with loops. You cannot say loop until we build a full permissioning system. I mean, you technically can, but I'm not doing it because I don't know which direction the AI is going to go. I don't know what features it's going to build. I don't know when…”
See the full analysis with timestamps →