On September 30, on Value-First AI Daily, Nico Lafakis took the question “which model is better” and answered it with a job.
“So the whole thing of which model is better finally has come to the point where the model being better is better at a specific thing … GPT will always be better. Pretty much will always be better at creating images than quad, especially if you’re trying to do something that’s fantasy, especially if you’re trying to do something that’s blended, it will more than likely always be better than GPT. I’m sorry, GPT will more than likely always be better than quad at doing that. Now, as it relates to vectorized art, quad will always destroy GPT. There’s a reason for this. GPT visually is based on pixels. Quad visually doesn’t exist. There is no image maker. There’s no video interpreter for Claude. So how does it handle doing this stuff programmatically? Because everything is made up of code at the end of the day.”
Nico Lafakis, Value-First AI Daily, September 30, 2026

Episode — Value-First AI Daily
Show and Tell: What We Have Been Building
The full conversation: Show and Tell, What We Have Been Building (September 30, 2026).
Open the episodeThe captions spell Claude “quad” in places. Look at the order of his answer. He gives GPT its win first, corrects himself mid-sentence to get it right, and only then gives Claude its win, with a reason: pixels on one side, code on the other. The claims are his, about those models, on that date. Every answer on this page is dated for the same reason: “best” has a shelf life, and the honest version says when it was true. What travels is the question. Better at what?
For HubSpot questions, Trisha finds Breeze’s answer is often better
The next day, on Another Orange Morning, a co-host asked Trisha Merriam whether she works alongside Breeze, HubSpot’s AI assistant, inside platforms.
“Claude is great. I love Claude and I use both Claude and Breeze every day. But, just for very HubSpot specific questions, I don’t find that Claude always has the answer. Breeze’s answer is often better.”
Trisha Merriam, Another Orange Morning, October 1, 2026
Her claim is narrow and hedged: “very HubSpot specific questions,” “always,” “often.” She uses both tools every day and says so before she makes the comparison. Breeze is not beyond doubt either. What goes wrong with AI-native work (opens in a new tab) has Chris’s account of the day it could not find a record until he pushed back.
AI will configure faster; a person draws out what the client needs
On October 28, on The AI-Native Shift: Beyond Industrial Defaults, Erin Wiggers answered what she took to be Trisha’s question about whether AI or a person will lead those conversations:
“And when I talk to clients, I’ve done this a lot. It may be the only time in their career that they’ve ever had to do something as big as a migration or a big integration. And they’re kind of flying by the seat of their pants, and having someone in the room to be able to say, I see this all the time, and empathize with where they’re at, I think is going to be increasingly the role of the human, where once we get to, okay, let’s go configure it, 100% AI is going to be quicker with that. But as far as getting the ideas out of their head and on to some kind of medium that we can all look at and kind of mold together, I definitely think humans have the edge there.”
Erin Wiggers, The AI-Native Shift: Beyond Industrial Defaults, October 28, 2026

Episode — The AI-Native Shift - Beyond Industrial Defaults
Why a Bad Process Becomes Bad Automation — AI Readiness, Part 4
The full conversation: Why a Bad Process Becomes Bad Automation (AI Readiness, Part 4).
Open the episodeShe names the human’s job, grants the machine its strength, “100% AI is going to be quicker with that,” and only then claims the edge: “humans have the edge there.” Her reason is a client who may be doing something as big as a migration for the only time in their career, “flying by the seat of their pants,” and who needs someone in the room who can say “I see this all the time.”
For trust, Chris points to George B. Thomas
On the August 24 Value-First AI Daily, Chris was talking about making Road to UNBOUND episodes with HubSpot insiders, and named the example he uses for trust:
“I love being the guy that just gets on the Road to Unbound episode with HubSpot insiders. Unbound insiders that see George as the rock star that he is. It’s larger than life like a brand that’s inside of the ecosystem and now they’re getting to create content with them and they have that level of reaction like, ‘My gosh, I get to create content with George Thomas.’ The memorability of it all, but the consistency that he’s had. If you look anywhere, you cannot find a version of George that was trying to pitch you or sell you … That has just built up an overwhelming amount of trust. So it’s very hard to do.”
Chris Carolan, Value-First AI Daily, August 24, 2026

Episode — Value-First AI Daily
Value-First AI Daily - Aug 24, 2026
The full conversation: Value-First AI Daily, August 24, 2026.
Open the episodeThat is a claim about a record, not a feeling. It is also the kind of best that cannot be claimed about yourself. It is built by consistency and credited by someone else.
When the best is out of reach, the choices are not only two
On June 16, on Another Orange Morning, the talk turned to finding a brand voice. Chris pointed to who does it really well, then to what most teams actually face:
“If you want to do it really well, George has so much content around brand voice. And collaborates a lot with Liz Moorhead on that, which I’ve actually worked with her before, and she’s just really good at what she does. But for most, I think, it can be a barrier because it’s like choosing between. My God. I gotta work on this for three months and get all the team together and figure out what our voice is. And you have so many arguments about simple words. Or, if I’m gonna create content, it’s gonna be AI slop, and it doesn’t have to be that way. Those are not your only choices, especially if you have never developed, brand voice documentation before. Just start creating and, as you use your own words, AI will tell you what the voice is.”
Chris Carolan, Another Orange Morning, June 16, 2026
He does not say to skip the best. He names George and Liz Moorhead for voice done really well. He says it is not the only choice, and that a team without three months to argue over simple words can start creating and let the tool describe the voice it already has.
Where the idea behind this page comes from
On June 18, on Value-First Partner, Chris named where he first saw “they ask, you answer” in action:
“I think I first really saw this in action from, you know, Marcus Sheridan. They ask, you answer … just because you give everybody the answers. Like, here’s the prompt template, here’s the agent, just here’s all the Markdown files. It doesn’t mean they’re all of a sudden gonna know and be able to use it to be successful. One out of ten maybe will, but that other nine are now going to know that you’re the person to go to get this stuff done.”
Chris Carolan, Value-First Partner, June 18, 2026
“One out of ten maybe will” is Chris’s estimate with a “maybe” on it, not a measurement. The mechanism is the claim: answers given away do not do the work for the reader, so the reader knows who to ask next.
Four of the answers here concede something before they claim something. Nico gives GPT its win first. Trisha loves Claude before she says Breeze is often better. Erin grants that AI will configure faster. Chris sends you to George and Liz Moorhead for voice done really well. The order is how you can tell an answer is about the job and not about the seller.
Worth passing on?

