Skip to content
ChatGPT · Codex · ClaudeEpisode 133 · 12 July 2026 · 01:08:52

The Problem With New AI Models Is No Longer Power, but the Person Who Has to Make Sense of Them All

Central question

Why is the main problem with new AI models no longer their power, but a person's ability to choose a tool and make a decision?

What you take away

Understand why the bottleneck is no longer raw model power, but tool selection, task framing, and the human ability to make a decision from the result.

Main threads

What to watch for

1Compare “Weekly Main AI News: ChatGPT 5.6, Fable 5 extended and Grok 4.5” with “Claude Design: what a new instrument can and who needs it”: they provide different criteria for judging the same issue.
2Test the conclusion from “Can the site be completely rewritten by Claude Design” in your own use case—what actually changes in the process and what remains a promise.
3Before choosing a product or approach, record the constraint identified in “New era of development and work with AI”.
4Define the owner of the outcome and the quality metric for the situation described in “How fast is the future of superintellant coming?”.
Signals to track afterwards
Watch for actions by Alibaba and Anthropic that confirm or challenge the episode’s central claims.
Compare new launches and policy changes with “Can the site be completely rewritten by Claude Design”: have access, quality, price, or constraints changed?
Check whether the scenario in “How fast is the future of superintellant coming?” becomes repeatable practice rather than a one-off demonstration.
Most useful for
DevelopersExecutives and managersAI usersProduct teamsTechnical leadersProduct founders

Key takeaways

00:00In a single week, the market received ChatGPT-5.6, a new voice model, extended access to Claude Fable 5, and Grok 4.5

The “Weekly Main AI News: ChatGPT 5.6, Fable 5 extended and Grok 4.5” topic becomes clearer once this point is included: this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.

02:49The practical meaning of the issue: chatGPT 5.6 Sol - first impressions

The practical meaning of “ChatGPT 5.6 Sol - first impressions” is that a benchmark measures a narrow capability; working value requires repeatability, a clear cost, and control over errors.

07:20Where the promise meets reality: cursor, Codex or Claude Code

The “Cursor, Codex or Claude Code?” issue should be assessed with one constraint in mind: a benchmark measures a narrow capability; working value requires repeatability, a clear cost, and control over errors.

11:13Grok 4.5 claimed victory in a parallel-development test, so the natural reaction is to add another tool beside Codex and Claude

The decision in “Claude Design: what a new instrument can and who needs it” depends on one criterion: this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.

13:38Claude Design shows another shift

The working conclusion from “Can the site be completely rewritten by Claude Design” is that this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.

18:32The market tests it through use: end of SaaS era

The “End of SaaS era?” scene leads to a working conclusion: the issue turns on whether the rule can be enforced and who carries responsibility, not merely on the existence of a new requirement.

20:52The boundary between value and constraint: china limits access to its best models

In the context of “China limits access to its best models,” this criterion applies: the practical boundary is defined by the agent’s permissions, the visibility of its actions, its action log, and the ability to stop execution.

24:54Who owns the outcome: grok 4.5 advantages over competitors

The decision in “Grok 4.5 advantages over competitors” depends on one criterion: the important signal is not one funding number: the next round, available runway, and closure rate show whether a company can survive the new cost of capital.

26:04China restricts access to the strongest models, companies debate whether employees may use Claude Code, and the market divides more sharply by infrastructure and geography

For the “New era of development and work with AI” scene, the decisive point is this: the new era of development provides enormous productivity while demanding an understanding of security, tokens, and data ownership.

01:05:59Voice AI can work in a car, help with tasks, and connect to the strongest models

The “How fast is the future of superintellant coming?” scene leads to a working conclusion: this section clarifies the mechanism behind the topic and preserves a constraint that would otherwise be lost in an overly simple conclusion.

What this episode is about

ChatGPT-5.6, a new Voice model, extended Fable 5 access, and Grok 4.5 arrived almost simultaneously. They can do more, but force users to choose modes, interfaces, and tools. A personal website case makes the shift clear: AI is already rewriting software, yet real value appears only beside a person who understands the task.

In a single week, the market received ChatGPT-5.6, a new voice model, extended access to Claude Fable 5, and Grok 4.5. On paper, this looks like another sharp leap. In practice, the problem has moved: powerful models are already plentiful, but it is becoming harder for a person to know which one to open and where it actually works.

Grok 4.5 claimed victory in a parallel-development test, so the natural reaction is to add another tool beside Codex and Claude. A closer look reveals that access and the working interface are limited. A dramatic test result and a product’s readiness for daily work turn out to be different things.

Claude Design shows another shift. AI can already help rewrite an entire website, move materials out of Notion, and replace part of a CMS or SaaS stack. But the model does not know which requirements actually matter to the owner until a person explains the architecture, content, and constraints. Generation speed does not replace understanding of the business.

China restricts access to the strongest models, companies debate whether employees may use Claude Code, and the market divides more sharply by infrastructure and geography. The new era of development provides enormous productivity while demanding an understanding of security, tokens, and data ownership.

Voice AI can work in a car, help with tasks, and connect to the strongest models. But it still mixes context, does not always hear correctly, and does not know when it should ask a clarifying question. We really are approaching systems that program software and chips.

The main advantage today belongs not to the person who simply has access to the most powerful model, but to the person who knows how to make decisions with it.

The advantage does not go to the person with access to the most powerful model. It goes to the person who can choose the tool, provide the context, and make decisions with it.

Episode transcript

The episode is in Russian; below is an English reading guide to the transcript (the full EN transcript is a machine translation). Voice matching applied to 92 segments: 56 identified, 10 mixed, 21 marked with ✓, and 5 unresolved.

Loading…