Transcript · January 1, 2026 · 11.3K words
Ep. 60: Fast Hours 2025 Wrapped: Tools, Shifts, and Wake Up Calls
Captions uploaded with the episode. Expect the odd misheard word. Tap any timestamp to hear it on YouTube, or go back to the episode for the summary, receipts and chapters.
0:00:00Episode 60 kickoff and end-of-year reflections
Play from 0:00:01 What's up, everybody? Midjourney Fast Hours podcast, episode 60 and the last one of the year. Drew Brucker, Rory Flynn, wrapping up 2025 in style. And yeah, man, we did 36 episodes this year. 36. It's like, I thought about that the other day. It's like, this is our not technically full second year, but second year of doing this.
Play from 0:00:29 And it's been awesome. Like, I first let's just get this one out of the way for all the for all the people listening out there killed it this year, like way more fun, way more people requesting, you know, or like sending us around and people, you know, getting seen the podcast. So thank you for spreading the word. Thank you for listening. And also thank you for, you know, engaging and like, just like interacting on this thing is drew. I, again, we don't know what we're doing. We're just talking, whatever.
Play from 0:00:56 like the other like what I was going to say too is it's it's this is great because one selfishly you and I get to connect and talk through this stuff because I think what I've learned and I'm sure you've learned too, because you're working from home. I'm working from home. You know, like this can also feel lonely even though you're working with teams and clients. You know, you're still working for yourself and um like just staying up to date on some of the stuff again hard on your own no matter what.
Play from 0:01:24 You might be saving things in this and that and the other. You can't do everything. So I think the fact that we had to come in here and talk and just kind of bullshit around is awesome, number one. Two, think like you said, it's been awesome to see other people chime in, comments uh in passing. Like I know you did some uh talks even internationally overseas and people coming up to you and saying they listen to the podcast.
0:01:50From niche experiment to mainstream behavior
Play from 0:01:50 I think. uh You know, we went into this without any expectation and I think we would probably still be doing this even if we had 10 people listening. But I think like that makes us feel good and it's always great when we see new faces too, because it's like, I had no idea you're even listening. So like those people keep popping up. So I think this has been super fun, think for like 2025, you we're going to recap that today.
Play from 0:02:20 A lot's changed, I'm just trying to rewind to where we were last year. think we were waiting for V7 to come out. um Video was still in kind of like uh an experimental sort of timeline in a lot of ways. Images, Midjourney was still king at V6. There was nothing that could touch V6 at that time in terms of realism.
Play from 0:02:45 uh just like, yeah, the way that I think you and I are going to touch on this too, but like the way the perception of this has kind of changed and people you were just mentioning this before we hit record, but it's just like. Less education on why this matters right? You and I were very early on this and a lot of people that listening to this were very early and if you were doing any kind of client work or just trying to tell your friends about it, whatever it is, it's just like you had to do so much educating and provide so much context on why this matters.
Play from 0:03:17 And now you don't have to, right? People and brands start to understand, like this has to be a part of what we're doing. And so that whole education aspect is sort of settled down a little bit. So we've seen a lot change since last year, and that's just kind of scratching the surface. nuts. remember when we were doing last year was, uh, we, we had like, I think it was runway act two, where I was, I did myself as Santa for your daughter to make her put onions in her socks or something.
Play from 0:03:47 Right. Like my cousin, and that would have been like, that seemed like, you know, for someone who wasn't in the know, that was like, what, what is happening? You know? And then this year, like my cousin came to Christmas and she was like, look, I made a picture on ChatGPT of Santa in our living room, putting presents under the tree to show our, you know, to show my daughter. I'm like, Like it's so mainstream now that someone that would never even consider that now can just type that into their phone and just like say, I need a picture of Santa in this living room. That's it.
0:04:00AI finally reaches non-technical families
Play from 0:04:13 Right. Right. more time doing something like that for my family, too. We were doing the Elf on the Shelf this year a little bit. Not seriously, but every couple of days we would have him doing something else and getting into trouble. But it was like uh I saw somebody on Instagram that I'm friends with that they're probably not in AI really hardcore, but they were doing Elf on the Shelf.
Play from 0:04:41 And they had shown something that they had basically done what you're talking about. They had taken, I guess, pictures of the house, right? And then had him insert it into scenes so you could almost document how he was being mischievous, right? That he was actually there. And I just thought that was a cool idea, too, at the time. like, yeah, that's, I wish I would have done that, because the kids would have loved it. But yeah, you're starting to see that filter down. Yeah. But that's it. That's like, you know, and then my uncle walked around and he goes, you guys know this is like what Rory does every day, right?
Play from 0:05:11 Like it was like a, it was like shout out because the shout out to my uncle who does follow on LinkedIn, who has no skin in the game and anything AI. He's just about retired and in the finance world, but he just appreciates the, the bander back and forth on LinkedIn. seen on there that I'm I thought that was your dad. My dad is also on there as well. They're, you know, that's what happens during retirement age. You get a little bit interested in what's going on. They miss, I think they miss it. I think they miss the work world.
Play from 0:05:38 I think they miss the little, I know my dad does. I know my dad gets into like modes where it's like, he's like bossing me around. Like he's still the boss at work. You know, and I'm like, Hey, you know, this is, this isn't like, you know, per, my last email, I'm like, why are you sending me an email like that? Or like, it's like, it's like, all right. dad, like, you know, I'm not going to schedule some time on your calendar. I'll call you later.
Play from 0:06:03 It's it's funny. They think they miss it, which makes me feel not that excited for retirement at some point. It just feels like kind of miss a little bit of work, which I do get. It's been a slow like two weeks with the Christmas. I'm like freaking out. I'm like, I got to do something. You know, like I need some chaos. yeah, it's like that right balance and you know, like sometimes those sprints are good for you mentally too, right? Because if you have too many of these sort of like lazy days, it's hard to kind of like pull yourself in.
0:06:18Why working solo in AI can feel isolating
Play from 0:06:33 Yeah, man, I mean, there's been so much shit that's changed in 2025. You know, I know you got a list we can pull that up to, which would be interesting. What about just Christmas in general? How is that for the fam? Did you watch any? Any good shows, anything like that? course. Chaos. Got uh shows wise, got Stranger Things in. um We got the finale coming out tonight.
Play from 0:06:59 Obviously doesn't even matter. I don't care. I'm just watching it. I'm having a ton of fun with it, regardless of how I think it is or not. It's just like event television at this point. Did Your Friends and Neighbors. um Finished that one up. Did the Diddy documentary, which was like throwing back to I honestly, outside of the Diddy stuff, I liked all the stuff on the biggie Tupac, like East Coast, West Coast stuff almost as much, if not more, than the Diddy documentary stuff, where I was like, I want more of that.
Play from 0:07:31 I want to go, I want to like, yeah, I want more of that from that perspective, because that to me was, who doesn't like 90s hip hop? Everyone likes it in some way, shape, form, because it feels like it resonates. It's storytelling. It's good. It's just good. music like it doesn't know the way to put it but It was still music back then. You know what I mean? was much more raw.
Play from 0:07:57 You had these guys creating beats on machines that, you know, like that's the artistry in my mind. And that's why I love it so much, the 90s hip hop thing. mean, and there were just, uh have you ever seen this? There's a really cool show now I think about it on Netflix. And I don't know if it's still on there, but it's a really good one where they go city to city and focus on the hip hop scene as it came up in each city.
Play from 0:08:30 So it's like it starts with, well, who really started that in Chicago or New York or Atlanta? And they tie it up, and this guy kind of finds these people and does interviews with them. And like, well, this is what I was thinking at the time, and blah, blah, blah. And then they. kind of go through this chronological order of how it evolved and like how more people picked it up and then who took the torch, right? And ah that series is great. I really, man, I gotta find the name of that, but I love that one.
Play from 0:08:58 That one's really cool because it just goes into the the grassroots nature of it. It's like to me, the music stuff is so fun, which I think we're going to have a huge, this is a prediction for 2026. Like I'm starting to see a lot more in the AI music space now. I think you're going to see a lot more of that get adapted because it is very mathematical in a sense. But like I saw recently that there was a new open source tool that came out and there's a few others, you you saw sort of like UDO get gobbled up by, think it was Warner Music, Suno is probably getting gobbled up by someone I forget, but you think like I just see that.
0:09:03Music, creativity, and early signs of AI music adoption
Play from 0:09:32 And then, you know, go back to like my, Tik Tok algorithm or my Instagram algorithm that that dude Ari on the street that I was sending you from the other day, the dude that just produces on the street and everyone in New York somehow has like, you know, bars ready for the it's insane. This is right. Just jump on. He's like, let me make a beat like for real, like in real time. And then bang someone just like ready to fire. I'm like, people are awesome.
Play from 0:09:59 I love, I like seeing like talented people like that. Just fire. You know, it's like it's all there and then they're ready to go at all times. But sorry for being off topic, people. But this is this is the end of the year. Yeah. I mean, so we last one we did was right before Christmas. So we're just kind of catching up with Christmas stuff. Then obviously, New Year's tonight, which is crazy. 2025. That's a wrap. Going into 2026.
Play from 0:10:25 lot to look forward to. uh I'm interested, know, obviously just like working in this space. I'm curious to see how that evolves. I'm also just kind of curious as a creator itself, right? Like just What new tools are we going to get? What is going to be the leap from now till this time next year? What are we going to see? And so, uh yeah, I don't know, man. Maybe this is good time to kind of bring up what we got for the 2025 stuff.
Play from 0:10:52 You kind of pulled a recap of some of those key moments in time from 2025. And this is not Midjourney exclusive, by the way. This is just more or less. Okay, yeah, so this is more like just the entire space, right? And so Rory put this together for us and then highlighted some of the key moments uh that he saw sort of chronologically through the year here. So what I did here for anyone also who's not, who hasn't been like, if you're looking for data polls, this is like a cool, this is like a, like a good sort of, I guess, tip.
0:11:02How fast AI actually shipped in 2025
Play from 0:11:27 Grok is excellent at querying Twitter. So like, can just like, instead of like using like the advanced search features, like using Grok to query Twitter to get like all this information. Cause really it has, like everyone just puts their model releases on there, but in regards to that, every other piece of information on there, it's great to just be like, searching you can it does look a fairly decent analysis of stuff, but that's just like a Instead of using like deep research on you know ChatGPT or Claude or whatever like if Like it really has like the bite-size information on Twitter to pull all the data um But that's a that's a sidebar, but what I did was basically ask it I'm missing some things here But I asked it for notable releases in the gen AI space so like image models video models the LLMs things like that just to like get an idea
0:12:14100+ major releases and why that number matters
Play from 0:12:15 Naturally, there's going to be some things that are missing here. But I just want to see how much stuff we got. And we got a lot of stuff. Yeah, like more than I remember. if you look at this whole list here from where at this is, god, where are we? 105 releases, which would technically, if we broke that down, be like almost every two and a half days, three days.
Play from 0:12:42 Right? So. And this is missing stuff. This isn't like everything. So there's probably some tools I'm forgetting, whatever. I did mostly like the heavy hitters. I did go in here and highlight, I thought, some inflection points. And I want to get your opinion on these. Because I know this is where I started to see some patterns where it was like, OK. Yeah. Yeah. So this one I probably should have highlighted was GPT 4.5.
0:13:01The real start of image editing workflows
Play from 0:13:10 Like to me, that was like their first notable update. in a while when we got 4.5 and that's when it started to get like way more GPT like I don't know if you remember this was like started to get in the that was like the new image model came out then we were getting like the you know the agent mode and like the codex and like all that stuff started like water falling after 4.5 yeah, I would say maybe even that right there.
Play from 0:13:35 4.5 was, that was the quote unquote release with the image model, correct? Yeah, because then for or for oh mini with the image generation, this is when it came. So it came a month after. massive though. I mean, just to recap that, we had no ability to take any existing image, third party image, do these little micro edits.
Play from 0:14:02 There was no possibility of that. All of those different use cases that we talk about with Nano, that stemmed from here. And I think that was a significant. moment for a lot of reasons. mean, like, if you're doing client work, I mean, this changed your workflow. And then also, if you're internal and you're paying attention to this stuff, this really kind of democratized some creative capabilities too, because you didn't have to be this savant in a particular tool.
Play from 0:14:32 uh Most people are using an LLM like ChatGBT, if not ChatGBT itself. So having that accessibility inside the tool for a lot of people I think was a really significant moment. I think March, we look back on this, probably the biggest shift of everything. Cause I look at when Reeve came out, Rev, whatever the hell we want to call it. Reeve was the first one that could really do text and style at the same time.
0:14:46March 2025 was the quiet inflection point
Play from 0:15:02 Like, whereas like ideogram could always do like text, but the style was kind of like whatever, like it wasn't good. Um, there's a few others that could do text, but it sucked in my opinion. Um, so like getting Reeve, getting Gemini and getting like GPT image all in that month and then having it be able to do text style and then know reef couldn't do the replication but Gemini 2.5 could so we had like because I remember I was sitting in Antwerp about to go on stage talking about like Laura models and stuff like that and I was like oh this is Laura's done like I don't really need to use it that much anymore because now I have these two tools so I was like Laura thing was a thing for a minute too, right?
Play from 0:15:43 It's like you have to create your own Lora's. And then the other part of this that you have highlighted too is that Flora AI piece, which sure you had comfy, but this felt like the more uh accessible, mass, sort of like appealing option for that kind of creative work, which I don't think even at that time, yeah, that note wasn't really getting enough hype, right?
0:16:06Multi-modal chat changed prompting forever
Play from 0:16:13 Even in March, that was not really talked about heavily, at least in social media circles, right? And I think that's where this space is really interesting, because I do think you can read the tea leaves pretty easily if you follow certain people or you're following a certain sort of echo chamber, because we are kind of. at the forefront of this. People like us, people like the listeners here, people like that are just providing this content of, look what you can do.
Play from 0:16:41 We're kind of actually running this whole push. It's a very democratized push, I think, from a creator standpoint. These tools aren't really having to educate anything. We're creating the content for them. So from that standpoint, Flora and that whole node-based thing was not even being talked about. It really wasn't. but it took Comfy and dumbed it down to like a very simplistic level.
Play from 0:17:07 And it wasn't even you could do as much as Comfy. You could just do basic things, but like seeing it, I think it was like, oh, this is what we need. Me personally, this is what I needed. I needed to be able to sort of like visualize everything in one space. But the other thing that was interesting here was the multimodal multimodal, if I say that right, um portion of like the GPT and Gemini, right? Like having that within a chat. talking to your images, being like add this, change this, do this.
Play from 0:17:34 Instead of it being a prompt interface, I think that's sort of like set things in different direction too, right? Yeah. of it too, with the coherence, just different, right? Before it's just like, we've become so used to sort of like this idea of a prompt being so much different than how we would talk, right? It would have to be formatted in like these little groups or by commas or, you weren't just talking to it. And I think that opened up a new avenue too, right?
Play from 0:18:02 Because I think the fact that, where we had been before that, you really kind of had to learn this language, you know, versus kind of now it feels a little bit more intuitive. And especially if you're already using LLMs because you were having to to kind of think in that way, right? Where it's almost like, especially on the talking mode, right? When you I mean, that's that's most of the time what I do now. If I if I have the time or I'm in the right place, I'm just talking to it, right? Because it just I can ramble. I don't have to worry about how I sound, but I'm also just kind of thinking on the fly.
Play from 0:18:30 I can pause. It's obviously quicker. I don't have to type it all up, worry about if I'm spelling everything right. So, I mean, yeah, I think that natural language aspect to your point was a really key moment. It's crazy to back and think about this because I remember that the day that these two came out because I was about to go on stage the next day and I was like, like everything I say today done tomorrow, you know, like kind of how it typically works whenever I go on stage, it seems like.
Play from 0:18:59 But that was that was pretty. Intriguing, I mean, I remember playing with it and I'm like, I'm going to put some of this stuff in the presentation like ASAP. And I was like, people didn't even know what it was yet. And I was like, it got released yesterday. It's really cool. Like, you know, but. As you can see, we'll keep pushing here. Then it seems like we slowed down a little bit in April. Doesn't seem to be much. mean, Cling 2.0 was a very good release. uh get Imogen 4, which was definitely good.
0:19:20Veo 3 and why video suddenly jumped ahead
Play from 0:19:26 Hylue, know, Minimax 2.0 was good. But then you see, I think, the tipping point of the whole year was VO3, right? Because this one, massive jump. This one took everything that had happened in the video gen space and flipped it on its head. Number one, we had the quality. Number two, you had the dialogue and the sound, which typically if you wanted to do that, you had to have four other tools. You had to do 11 labs to create the voice.
Play from 0:19:55 You had to do something to have a lip sync. You had to do something to make the background noise, whether that was like MM Audio or Borrello or whatever it was at the time. All in one tool. Then you're like, oh damn, this is good. Nothing else was close at the time. So I think that was like the, like everyone needs to push the gas pedal now. just sort of threw gasoline out the fire.
Play from 0:20:21 And it's like, yeah, yeah, yeah. the crown here because I mean, right? Up until 2025, a lot of just, it was so quiet. I mean, to the point where you almost started to worry about Google. You're just like, this, are we gonna see the disruption, like the fall of a giant in real time? Because it's like they got all the resources, right?
Play from 0:20:46 They've got the money, they've got the talent, they've got... you know, the ability to designate entire teams to figuring this out. And it's just like nothing's happening. And then all of a sudden, 2025, I mean, bang, bang, bang, bang, bang. mean, almost a big jump for them every month in some way, you could argue, right? Like, there's probably some things like, I don't know, I'm just thinking about like Pompeli, for example. I don't know if that's on this list, but like that idea itself is sort of like a game changing thing, right?
Play from 0:21:17 It's like... we finally have something that can actually start to get into that space. And it's obviously early, but they had all these different pillars that they kind of knocked down. The VO3 thing, I mean, they're still winning video. mean, 3 holds up. And we were in December. We were in December here. I mean, so that's seven months ago, six, seven months ago. That's going to open up a broader conversation when we get to the end of this, which I think is going to be, you know, where, what the difference has been, like for the both sort of mediums.
0:21:41Why Google quietly dominated 2025
Play from 0:21:50 Cause when I keep scrolling here, I feel like that's the inflection point, but then like, it really doesn't get that much crazier after that. Like Weevy, I put Weevy in here. Cause again, I think that was a pretty tipping point for myself when I was like, God, like systems batch processing. It's easy now. I don't have to like, engineer, like I can actually just visualize and get stuff to where I want to be, right? Because then we have, you know, a bunch of other image models coming out, Grok 4, Higgs Field, Sol, you know, the Wands, the Quens.
Play from 0:22:21 Runway of Left, I thought, was a pretty interesting sort of update in July when that was basically video-to-video editing, which I thought that was sort of like the next, sort of like the Nano Banana for, you know, for a runway or for a video. And It wasn't, you know, it had its faults. It wasn't great, but it's sort of like set again, as Runway typically does to give them their credit. They typically do things in a way that like everyone else is like, I should probably do that too. And like then, you know, see other models that have that like cascading after the fact, but they were the first, of course.
Play from 0:22:53 So like that to me was pretty, pretty impressive because then we get down to where is it here? um Maybe I lost it. Yeah. go down a little bit, there's online, 51 there, September 18th, mean, LumaRay 3 drops. That was like this huge sort of like cinematic upgrade to video and that hype lasted 48 hours.
0:23:00Why hype cycles now last 48 hours
Play from 0:23:22 48 hours. point and cling 2.5 great that was another one I don't know why I don't have Gemini do I have Gemini on here? I don't know what happened to I don't know what happened to Nano Banana Somewhere in here. I lost it, but we got we got Gemini right around Labor Day which was around the end of end of August into September which that was like the You know that was really like another kickoff point if I'm looking at the three kickoff points this year You know like it's been Google like it's been sort of Gemini 2.5, it's been VO3, it's been Nano Banana, and then Nano Banana 2.
0:23:51Nano Banana Pro and precision image control
Play from 0:23:57 all, like to me, these are me personally. Like I just see that as being like the industry shifting for each one of those. Versus like here's another update that fits the sort of mold, if that makes sense. But you know, that's sort of what happened. had Sora come out. Yeah, Claude. And the LLM space, we need like two episodes to do all the. sore two another good one.
Play from 0:24:22 Because it felt like that was a huge leap in terms of like social meme content. You know, obviously like like broader awareness to like, not the normies, but like the world. Yeah, so it's like anyone could use it, anyone could have fun with it. ah It's not hard. And it also did like one functionality that you saw in there was the ability to structure like full story narrative in that without having to prompt for it.
Play from 0:24:53 So create me a trailer and it would do the cuts and the different scenes and think while it was doing it, right? So. Boy, it's gonna be really interesting to see what happens at OpenAI in 2026. I mean, obviously they've just got endless. They've got the infinite glitch money code over there, but I feel like people are moving away from them. I mean, not like in a mass, but myself included, it's just like, I want something better.
Play from 0:25:26 The LLM is still doing things that should have not be doing now. you know, that Claude can do. You Claude is much better at writing. I just like the things that they have set up, Claude, with you can easily build these artifacts and it actually listens. I mean, you try to do that with a custom GPT, it ain't fucking doing it, you know? um You know, or if it does, if it can, in some cases, it's it's harder to achieve it, right? And I just think, like, that feels like that's something that OpenAI should have figured out by now.
Play from 0:25:55 um it's interesting with them. I don't know, because everything else has gotten good. know, like Gemini, it's great. Grok is great. Yeah. really interested to see what happens with GROK in 2026. I mean, they made a lot of steam in the last six months here. If you've looked at the condensed nature of who's developed the most and probably the most jumps, it's probably Grok in the shortest period of time from having like having nothing to being like pre I use Grok for anything that's query based now, like for deep research.
0:26:02Grok as a real-time research engine
Play from 0:26:29 It to me is the best. even I don't know what the information like at least I know like the information's coming from X, which can that could be good or bad, but it's really solid at generating or like queries from like even going up and looking like It's got that in real time, it's got that in real time connection that we're all wanting that chat GBT you're not sure when it's going to give you the most real time or not.
Play from 0:26:57 Like it's not super reliable always. You might have to double check your work. Whereas you feel like that's also been the benefit of X or Twitter in the long run. Anyways, like for the longest time, that's always been the place where breaking news happens, you know? And that's what I've always loved about is that real time aspect of If I'm not watching a game, can follow along with people that are watching the game. know exactly what's going on. You want to judge public sentiment to like, that's the easiest way to, to like analyze data.
Play from 0:27:24 You don't have to go and like jump through hoops to pull posts into like things and be like, can you read this? Oh, sorry. need to do authentication from whatever. It's like, can just query like here, take this post that's viral and like, give me the sentiment analysis on all of comments. And it's like, 33 % like this 80 % and you know, 70 % hate it, whatever the hell the numbers are. My math just sucked. But like, there's so many cool ways to use it because of that function. Um, but One last thing on the video too, was the physics, right?
0:27:49Why physics in AI video finally started working
Play from 0:27:55 This time last year, physics were still F'd. And I feel like that in a lot of ways made a huge, huge leap forward. I'm talking about like the real time physics of human bodies, moving, walking, running, flipping, jumping, all that stuff. I mean, you couldn't do any of that stuff in 2024. No, butch, which is like, again, we started moving through here. have figma. I have the figma, we be acquisition as a big one because I think that sort of set.
Play from 0:28:23 I think that sort of set again in motion that this stuff is real and it's useful and you're going to see a lot more companies just acquiring. We just had meta acquire Manus yesterday or two days ago, which, you know, another agent building tool. You can feel like the acquisition stuff is coming now. Like all these all these other companies have like what the smaller you know, startups like incubate and develop and gain follower base and like go through a bunch of failures and now they have good solid user base.
Play from 0:28:50 And then it's like, let's just die them now. So we don't have to develop it for two years and spend a ton of money. But you know, you saw that whole thing where we had like all the node based stuff happen like real quick. And it was, it was, that was November, October. Now everyone's got nodes and that seems to be like the most comfortable way to create. And you saw a lot of people transition. I think to that, which was to me pretty interesting because it felt, think people just felt comfortable with nodes because either they'd used them before in the past or it's just like an easier way to visualize first typing into the prompt and pray box.
0:29:12Nodes, workflows, and why visualization matters
Play from 0:29:25 Hope that something comes out. So that was, you know, that was to me where I saw probably the last of this, where it was like, okay, then we got Nano Banana and then we got, know, cling had his updates, which I thought were great. I didn't think it was like revolutionary. I just think it's a really nice. quality update to their platform. Flux 2, I'm like meh on it. I don't think it's exception. It's not Nano Banana 2.
Play from 0:29:51 ChatGPT, the new one in 1.5, I don't think it's as good as Nano Banana. That's just my opinion. um about seed dream a lot too. I don't know if they just pump it. I mean, it's fine. It's not Nano Banana. not. It's still not as precise. Like, Nano is precise. Like, it's very precise. um think selective realism can be similar, but you know, if you were to run that experiment over the course of a various number of images and different types of images, Nano is going to win.
Play from 0:30:24 It creates, it's game. It was like AGI for images. I think that that that was for me top. That's a top two or three jump on the year because Nano Banana was also was was obviously great, but the jump in the resolution in detail and then I was running into that problem, which I've talked about on this pocket right where it's like I want to do.
0:30:26Why Nano Banana Pro felt like “AGI for images”
Play from 0:30:50 Let's just say I want to do a flat lay of all these ingredients, multiple ingredients. I also want to get text and arrows in there and blah, blah, blah. You couldn't do that until pro, right? So like that advanced creative control aspect wasn't present. And then, you know, the editing capability. um And I would just say like overall realism, like you pump in Nano Banana Pro. uh There are photos basically indistinguishable from reality that we've got there with that, right?
Play from 0:31:22 you know, this has been. Man, what a uh year. What a year. uh Will we get as many updates in 2026? Yes or no, and why, you think? Yes, double time. But this is for everyone here, just so you know we're not effing crazy when we talk about this stuff and being like there's something new every week. I just pulled that data into Clawed and said make a visual chart, right?
0:31:26Will 2026 move even faster?
Play from 0:31:50 And you can see here, like how many releases we're getting a month. So it looks like we peaked in April. And again, companies pumping in for Q1. So we're thinking about this also, too, Rory and I, from the aspect of VC money. So these companies that are funding them, they have certain deadlines and goals they're trying to hit that are based on quarterly, uh yearly calendar. So the end of March is important.
Play from 0:32:16 Or depending on when you're counting, it could be the end of April. But you see that here. And then, you know, like, like you said, with this chart, you know, kind of like Q2 summerish kind of stuff, but then Q3 and Q4 just go. bang. So when we said like the last couple episodes and it's just like, there's something new every other day, we had 15 major releases from the major companies, which is every other day. So it's, you know, like, and this isn't even counting, like half the stuff.
0:32:25Release cadence, VC pressure, and reality checks
Play from 0:32:43 This is just like the major companies that I put in there. I didn't even, you know, there's a lot of other micro tools that did not make this list or tools that people probably use. put anything about Canva stuff in there or Adobe or like, you know, Like there's a bunch of things in it. This is just from like the major AI players, the Googles, open AI is, you know, Alibaba, uh, bite dance, things like that. And like, it's crazy. It's like, so you see this curve.
Play from 0:33:09 I think this is what we have to get used to for next year. This just maybe I just have a feeling because that's, it kind of felt like this ever since last year, the end of last year, right? Where it's like, can't keep up anymore. was, that was sort of like the idea. we were to run that through, like we didn't compare that to the year before, obviously, but yeah, I would agree. It did get exponentially uh more frequent with the updates, more companies involved.
Play from 0:33:39 I wonder if there would be any reason that it wouldn't like again, compound upward and to the right. I would just say maybe it's just because we get some consolidation. Maybe. probably it. But otherwise, the pace, theoretically, should be just as fast, if not faster. Yeah, which this is I don't know why this was eye opening to I feel like I just wanted to validate. Validate that we weren't nuts.
0:34:03Images vs video: who’s actually ahead
Play from 0:34:06 Yeah. Yeah. Yeah. to look at and say, that makes sense, right? Yeah, because we were just like, I remember us like everyone's probably like, all right, dude, enough complaining. But it was it was probably during like October, November ish, where it's like, how do we even how do we even look at this stuff? Can't even tell you what's going on with it. Because by the time I tried, you know, cling two point five, there was Luma Ray three and then Vio three point one and then whatever.
Play from 0:34:35 it was right. But that's the that's the one thing that I want to say over all this. I want to like talk about that is that the I feel like images this year went from like here to like like if we're climbing Mount Everest they went to like Base Camp 2 you know like it's almost close to the peak like I feel like video sort of just like like we got to VO3 and then like nothing really got that much better after that like and there's still like images you can do so much with and video there's still consistency issues physics issues like you can't hold consistency through a number of different things.
Play from 0:35:14 There's still only 10 second generations in most cases. As the max, there's still issues with sound and dialogue and consistency across those images. if you're going to do key frames and extending, there's pacing issues. There's a lot of issues. I mean, there's a lot more variables with video. images seems to have really... Remember when we were talking about this with Midjourney? It's like, much more photo realistic can it get?
Play from 0:35:40 How much... There was like... That's right. a lot we can do now. So do you feel that way? Or am I just nuts? No, I do feel that way. Dude, I think I think we're going to see some stuff from Grok in 2026. I think they could be the sleeper in this because. Grok 5 is going to come in Q1, I think. Well, well, it's definitely come in Q1.
Play from 0:36:06 I guess the question is, is it coming in like January, February? You know? em Because they've like to to both of our points where we're talking about, I mean, they've just gained so much ground so quick. And I mean that what they're doing with their image and video right now. I you can make a case like they probably have the most approachable video at your fingertips. Do something with it right in the moment than anybody.
0:36:18Why Grok might be the sleeper winner
Play from 0:36:34 You're already using social and now you're just it's right there in your feed. Is there any other way? No one like I haven't been on Facebook, but you know, but like Instagram they don't have anything like that where you can instantly sort of like create while you're. scrolling. This is a new thing. Now that I'm talking out loud about this, mean, the fact that you're able to create while you're in social.
Play from 0:36:59 it's, yeah, okay, yeah, okay, good point, good point. But it's just like, now people, if I wanted to roast somebody right now, I could find, let's just say one of these online trolls is talking shit. and they've got a picture somewhere of them on their profile. I mean, you can go to Grok and say, put this guy in a clown suit now.
Play from 0:37:25 You know what I mean? And they'll instantly generate a picture of this dude in a clown suit. You know what mean? Like the fact that it's just that uh easily accessible and low barrier to entry to do something. I feel like that on a mainstream social platform has been sort of like this silent thing that maybe we haven't talked enough about. just as a group, as a collective.
Play from 0:37:50 So I'm really interested to see what Grok does, man. I mean, I think like we're also probably gonna get those longer watchable videos. Obviously like that stuff's gonna happen. uh I wonder if Grok, I wonder if they'll get into sort of like this workflow component in it. I don't know. really, I think it's a good point though, dude. I think it's a really good point about Croc. Like if you look at it, it's just like you see who the players that have risen to the top have been, who has the most data, right?
Play from 0:38:20 So you have, you you have YouTube, you have Meta, you have Google. I mean, sorry, YouTube is Google. You have ByteDance, right? Like those are the ones that have sort of really like pumped it because they have the data and we're just feeding data into it. X and YouTube and you know Instagram and whatever at all and TikTok at all times so it's like they seem like the the ones that are most prime but I it makes it like makes you think back to the to the acquisition of X and it's just like was 45 bill or whatever Elon paid for it was it underpriced did he get a buy there did he get like a deal for what Grok's about to do did he know he and he knew
0:38:36Data, platforms, and why distribution matters
Play from 0:39:05 Well, dude, I mean, oh let's not get too political with this, but the fact that he bought it anyway and you had somebody that wasn't going to censor a platform was interesting. And that was like the start of this. So if that doesn't happen to begin with, which was a real reason why he probably purchased, or most of the reason, where are we? You know, I mean, it's just like it's an alternative universe that I'm sure is probably happening in real time right now that we're just we're not we're not on that field, but.
Play from 0:39:38 But you see it though. Like you see now, like he was a founding member of OpenAI. He had to know data is what feeds this. He had to know that X is a... he's already busy as F and then to get involved in them just, man, just putting on it like a literal jet pack. No pun intended. They're doing all this other rockets. I Tesla, mean like all this other stuff. And then you're just boom, like there's this whole side.
Play from 0:40:07 him. We're not here to grease Elon. So if anyone's gonna get upset about that, we're just saying like, that's an interesting development. Yeah. thing, right? Like if that does frustrate you anyway, I would ask you to think about like this. There can be good and bad parts of anybody. Right? And this doesn't have to be a I like them or I hate them thing. You could think there's a whole spectrum here. So keep that in mind. But I think the fact that you've got somebody that's so like vocal and then involved with this and now.
Play from 0:40:39 rock is just supercharged into this conversation is interesting. Man, I'm really now that we're talking about out loud, I might be the most interested, maybe with the exception of Google to see what they're going to do. And maybe more than Google. I don't know. He knows what to do, obviously. He knows how to, he knows the inner workings of OpenAI.
Play from 0:41:05 He knows how it was built. I use Grok a lot more than I thought I would. Not from the image side, but just on the LLM side. It is little angry in nature though. It's a little bit, which I kind of appreciate. It's got a little bit of an attitude to it. I'm like, all right, that's cool. It's definitely got something you know, you say it now. That's a good prediction for 2026.
0:41:28Consolidation and acquisitions are coming
Play from 0:41:31 I do like that one. That grok is probably going to be is the most unbiased LLM. I could see it. It's got the most variance of opinion. Yeah. It has to have the most variance of opinion verse anything else. Because, you know, everyone else is sort of like naturally more censored. So there's only certain things I can get through. Right. But it's, you know, Grok is Grok is slowly becoming on my on my research side of things.
Play from 0:42:01 The number of the place I go to first. And yeah. But. use it from time to time, but I think it's one of those now that I should be using daily. I still go to ChatGPT for things just because I have a quick shortcut on my phone, but it's like, probably should be Grok. So. is overly verbose. Like I have to go in there and do the custom instructions to trim it down because it'll write me like a novel, which I'm like, OK, cool.
Play from 0:42:26 But whatever. It's like I don't need all that. uh But dude, it's you see it like I see it with the tick tock stuff. You know, it's it's like getting way more like the bite dance stuff and everything like that. It's getting more uh pressure by dance stick talk. I'm forgetting that right now. But it's the YouTube video three. Right. Like bite dance, C dream, grok, grok image.
Play from 0:42:52 Like there's so much happening in that space. There's no way to like, that's it. Like, I just see that being like the trajectory upward. Like, I don't know who else meta gets manas like who else is going to win? I don't know. I don't know. Adobe's not like I don't you know, I like Adobe, but they seem to be just like, all right, fine, we're go the partnership route.
Play from 0:43:19 because they can't keep up. They have to do the same thing. They should probably start acquiring people. it's... The consolidation's coming. Yeah, consolidation's coming. maybe that's Reeve. I don't know enough about Reeve in terms of like who's behind it, how it's funded. But that feels like a good candidate if you are wanting to acquire a reputable model that does good work.
Play from 0:43:48 The foundation is laid. Minimax, I would feel the same way about them. they have some, they have really good model. I feel like, I feel like Midjourney is probably doing like some, Midjourney feels like they're working with a Chinese model for, for something. You know, whether it's, whether it is Minimax or Kling or someone, I don't know. It feels like would Seadream, ByteDance, whatever.
0:44:14What Midjourney must do next
Play from 0:44:15 about, okay, let's talk about mid-jury for a second. So VA should be January. I mean, it's hard to say because we haven't gotten an update on office hours in a few weeks. So I don't know if that timeline is still working, but let's just assume you're gonna get a February at the latest. I think we can probably feel pretty confident about that. um I've asked you this before, let's look at this maybe a little bit over the course of, as in fashion that we're doing here, over the course of 2026.
Play from 0:44:48 What does Midjourney, what does their path look like in the next year? And this is a little bit tricky, right? Because... There's stuff going on behind the scenes that we have no idea what actually is happening. But I think this is the year that we're gonna find out exactly where Midjourney fits in this puzzle. I guess, what are you thinking they should do or have to do or are going to do over the course of 2026?
0:45:23Image editing as the make-or-break feature
Play from 0:45:24 Image editor. That's it. That's it. Like it has to be there. It's it. That's all. I mean, just to stay relevant and competitive. I know they they have bigger plans than that, but I think that needs to be. I've said it once, I've said on this podcast, everyone's probably like, dude, we've heard this enough from you, but like, that's it. Like, that's just to stay. That doesn't mean like pushing anything else forward. That's just like to stay current.
Play from 0:45:51 Like the other thing is. And hey, maybe one, I was just gonna say one other one. No, go ahead. You go ahead. I'll jump in after that. last thing is with some of my most like visually interesting Midjourney images I can't take them into the other tools they still don't work so like nano good but doesn't get there seedream good doesn't get there like tell me what you mean in terms of like the prompt itself. If I take something like this into one of these tools, like it's very hard for it to mimic sort of like any of these things, because some of this stuff is just too intricate for it, right?
Play from 0:46:29 Like this is like I've had a hard time getting this one to work because of the intricacies of it. to like if you were to create an image in Nano doing it? Are you saying like it's hard to do in the sense of like recreation or just in general of the level of detail just across? say I wanted a different view of this exact character. I've been able to do that, but it's just not the same vibe.
Play from 0:47:01 Let's see if I can actually pull up the example of it. Where I have, let's see, sign in. Let's pull this up. I was trying this with, maybe it's this one. So was using this guy, The one that we just saw. And I got some good stuff from C Dream. I was trying C Dream. I was like, let's just check this out, see what it can do. But it's good.
Play from 0:47:28 just doesn't, it feels flat. Yeah, I don't see, like this one maybe, I'd be like, that's cool. It still feels like it doesn't get the Midjourney vibe on this. ah You know, like it still feels it doesn't feel like mid-drain. Yeah, it might be doing the it might be doing this, but like I'm not not really like overly impressed with it outside of that.
Play from 0:47:54 If I go look for the let me see if this one is what I'm looking for where I had it. In here. This one, yeah, so these are more like the nano, but I was I was kind of vibing with these, even though it wasn't the aesthetic, I was trying to put it into like the you know, that flash photo sort of look. And it's cool. I mean, it's grungy. It's sort of raw, right?
Play from 0:48:20 Like it looks the way that I'm asking it to, but it's not necessarily exactly what I'm looking for. Like I still want to keep this sort of like color palette and this, this very like soft shading on the edges and like all these other things that I'm trying to ask for, like this really like rich, good color, clean. Like, dude, I was having a ton of fun. with this one yesterday, just going back and forth.
0:48:43Workflow fatigue and creative burnout
Play from 0:48:46 Yeah. you could mess around with that for a while and get it. The problem is it's still a pain in the ass to do. That's where I think the workflows have started to show up and become visible. And we've gotten these capabilities. Obviously, nodes is a great example of that. But I still think there are all these disjointed pieces along the way still, right? Which we've always talked about historically with Midjourney as the example, right?
Play from 0:49:14 five steps that now today only take one or two. I think it's that part for the workflows too, right? Like you know what you want, but you're, and you could get it. It's just eating up more time and energy than it has to. And that becomes, you know, more of a pain in the ass than it is fun. uh And obviously like you're probably using this too for, you know, well you are for client work, but that's where it's like.
Play from 0:49:39 Well, all of sudden, everything speeds up if we can start to fix some of these disjointed pieces along the way of the workflow. There still are those pieces. Even when you're going into Weevy, I know you've probably got some templatized system prompts, but I'm just thinking, in general, that could take a while. That's a piece. Then the prompt is, that's a piece. all the connections, those are pieces that ideally get consolidated somewhere in the future here.
Play from 0:50:08 And I think like maybe we start to see that a little bit too in 2026, the workflow consolidation, you know? Yeah, I hope so. Like, that's why I'm trying to build like little micro tools and things that just like get rid of that for me. Because again, I just want to be able to the way that I iterate this quickly on here where it's like, you know, hit remix, go, enter, you know, blah, blah, blah, Like getting to the one image is is relatively straightforward.
Play from 0:50:34 But then like taking this person now, I want to do something with them. You know, like that's where I just want that's where I want Midjourney to step up. It's just like okay. I like this. I want to play with this. You know, not just like, that's it. One vacuum done, you know. there we go. I think that's, if I'm Midjourney or just speaking for myself and it sounds like you too, that is a top two or three in my mind is if you can kind of start to consolidate the storytelling process, right?
Play from 0:51:10 And I think like the example I used to give like a while back was like this idea of like, hey, if I've got an idea in mind, like why can't I generate three or four images of that scene instantly of the story, know, like almost like a combined LLM format, right? um You know, just again, to like really quick, activate me to create that. Otherwise it's just all these single pieces, these isolated ad hoc things, they're they're exhausting.
Play from 0:51:37 Yeah, that's my problem. I want to play with this stuff more, but now I have to go bring it into a different tool. And then I have to run it with a different system prompt or make sure that I'm deconstructing it a little bit so that aesthetic is exactly what I want. Meanwhile, Midjourney just knows what I want. And if it could just do it, which I think it will be able to. I think they've mentioned that. hey, you know what? That's also an interesting thing, which is Midjourney knows what I want.
Play from 0:52:05 Mid-journey has that. And I think it's possible to kind of do this, right? Like so for even in free pick, for example, you could set up a style reference now and kind of do like the mood board thing that you can kind of do in Midjourney. I haven't played with it enough to know how reliable it is, but Midjourney knows what you want. That is still a competitive advantage. And I'm really interested to see now too, that you brought that up, the personalization, the next evolution of the personalization, because up until now, there is no real precedent here with how to personalize what you want outside of just like this mood board.
Play from 0:52:42 Let's just call it, know, MJ had the mood board. You've got a style reference, right? Like some of these platforms now have that Midjourney did that. Now what's the next iteration? Well, I think Midjourney is going to release that next iteration. I think that that's a pathway where I think they're ahead that they could potentially impact these rest of these tools by the way they do it. So really interested to see how that personalization comes. And then the twofold part of that that you bring up is just that saves you a lot of time.
0:52:50Personalization, mood boards, and creative joy
Play from 0:53:10 You know, if you've got a model that knows exactly what you want, you don't have to think quite as hard. You don't have to be so contextual on what you're doing. And that's just one, you know, extra step that you no longer have to really like pour in a lot of energy and time into. And I just think like that's the really important part of this is consolidation, less exhausting, more fun, less pain. Less pain. Yes. And just like, this is, this can be fun.
Play from 0:53:38 you know, this is, I know David always says, you just want to make beautiful things. It's like, yeah, this is, this is beautiful to me. I want to do more with it. I want to make more beautiful things from this one beautiful thing. Right? Like that's, that's sort of how I, like, I don't even need to create anything for work. Like, I just want to play with things. Like if this is going to be, if Midjourney is going to like, keep staying as the, you know, the artistic rendition of AI where it's like, I want to do.
Play from 0:54:03 like things that just like express myself, then like I just still want to like express myself. Like I don't even need it for doing client work or project work now that that ship has sailed. Right. I don't need it for that. But like for my own part, when I'm doing this yesterday, just like Midjourney hooks me back in. I get an idea that I like, you know, you just start hitting the enter button like a million times. Like I want to keep going. Like that's but then I stop because I'm like, let me take this into nano.
Play from 0:54:31 it doesn't work. And then I go back and then like all. lost, right? I'm like, this is sort of like I wanted to do more things with this, but now I don't feel like thinking that much to do, you know, to like, like, I know what I want to do, but I don't feel like going to engineer it somewhere else. I know I can get there. And I know I can do it. It's just like, I'm playing like this is this is playtime for me. And maybe it's not for a lot of other people. But for me, this is how I use it.
Play from 0:54:56 Like, I'm not getting, you know, these images from seven style references. And you know, an assist from Marco on his, you know, his crazy token ChatGPT or is it crazy token GPT? When I'm like, dude, put this image in here and tell me what it spits out. Then like, we're like co-collaborating on this. And I'm like, sick. Let's play with the parameters. Let's change up the style reference or the style weight, the, you know, stylized values, throwing the parameters or the permutations.
Play from 0:55:24 So like, it's the only place I can do this. And then I just don't want to go anywhere. Like I just like it here. If you guys are listening, I like it here. Just let me stay here. I, yeah, I love that one. I also just think about like that whole mood board thing. It is really good. There is still room to take that further. Totally.
Play from 0:55:50 Totally. kind of to your point of what you're doing and what a little bit of what I've done too, in like that node side on Weevy, right? Where you can kind of start to get into more of an exploratory mood board and it's not so constrained to what you put in. So I think like if there's almost like a prescriptive way to kind of put things in there to what you might want, like an ideation partner in that sense visually.
Play from 0:56:18 could be really interesting to iterate on. like, that's another path, you know, like that could be really interesting for Midjourney is kind of like being a little bit prescriptive and almost like giving you things that you might envision yourself needing, right? Like serving it up for you, serving up new ideas for you, which has kind of been the way that they've kind of done the explore page anyway, they're serving up ideas for you.
0:56:44Why mood boards drove the best work of 2025
Play from 0:56:45 That's, I looked back at all my favorite images from 2025 and I would say 90 % of them came from mood boards like that I built. it's like, it's probably the function that I use the most and like made me, made me the happiest within the tool. Cause it's just like, yeah, that's what I'm looking for. Like I want that. And fun one for anyone else who's done this exercise. If you go back and look through some of your favorite images from the year, throw them all in a mood board.
Play from 0:57:14 Just like let that be your new aesthetic like it's a different different way to like sort of optimize personalization in that sense because like this one This one mood board I have which is things called vibes or here something I just put like random images that I liked in here. I don't know where the hell it is If I can find it yeah, this was just like Random images that I had liked from either like the month that I was creating or some of my favorites from last year Which there's definitely a lot of v6 personalization, like a personalization mood board if you.
Play from 0:57:46 And this one tends to work really well on every on every image that I like, because it's just like sort of the aesthetic that I'm going for. None of these have any sort of rhyme or reason to them. There's images that I really like that I've created on the platform. Right. And it's like, OK, let it ride. And then, you know, that's what's that's what's going in on here is on these mood boards. This is it's in there on vibe. In addition to, you know.
Play from 0:58:12 Like I've been doing that instead of my global profile. Like to me it's been working better than my global, because God knows what's going on with that thing. But... torched. I don't even mess with the global no more. Ever since I did that V6 thing, like, I usually combine, if I'm not using a mood board, I'm combining like my V6 personalization code. And there was one other personalization I have that's just around like photo realism specific.
Play from 0:58:38 And I just stack those and it's worked pretty well for me. But I think it's interesting what you said about the mood boards, being a lot of them came from a mood board. think that's. Probably true for me too. you know, that's even with me taking a huge gap. Like I took a long break from mood boards too. We went in on them. When did those come out? It's like probably first quarter or either. Oh, you know what?
Play from 0:59:04 It was either December. I think it was around this time last year or January. mood boards. Because January we have here blending mood boards, so it definitely came out before then. It was probably December. OK, so Q4 last year. But yeah, I mean, we went hard on those for the first month, you and I. I mean, we really did. We put together those whole mood board packages. And then I probably stopped for nine, eight or nine months. And then the last I would say Q4 for me, like probably since, yeah, September, October, I'm right back.
0:59:12Personalization profiles vs mood boards
Play from 0:59:36 I've created some really good ones that I just pretty much using for all my creative stuff, you know, and I just you're supplementing your personalization profile because you don't like it that much, right? Like that's how I'm doing it. It's like that and my V6 personalization profiles versus the V7. That's how I've been getting what I like again. And it's like, once it brought me back, but like I now have more fun on there because I've been doing it that way. Because there's something about without personalization on like the bass model like doesn't interest me.
Play from 1:00:07 And then when I do put it on, it just alters it too much. So I'm like, I need to a happy balance of this. that seems to be it. There's also one other thing about Midjourney that you don't have, I don't think with any other platform that I can think of, which is there are so many different ways to still get good results. There's just not that many variables on a lot of other platforms.
Play from 1:00:33 And I think in some ways it's good to have less variables, but I think from an exploratory standpoint, I always wonder what am I missing by not putting, you by putting chaos five instead of 20, or putting this in a 16 by nine instead of four by five, or using this mood board, or this mood board, or stacking mood boards, or stacking SRFs, or stacking personalization.
1:00:43Why Midjourney still feels different
Play from 1:00:58 There's so many possible outcomes. And again, that goes back to the personalization of the model, which already makes it more enjoyable. I'm not shooting in the dark ever. mean, you are, but this model knows you. And when you have that many, I guess, creative ways to get to good outcomes, that's pretty cool.