Marching To The Beat Of Your Own Drum

It's been an interesting six weeks since I started down this path. What path? Six weeks ago, I began a journey towards starting a company. A multimedia company I never knew I had always wanted to start before my introduction to AI. You read that right. For thirty years, I had toiled like so many before me, working on my craft in the hope to share with others. The old saying is that an artist's life is one of suffering. Not only to create, but also to be heard, read, or seen. 

GPT-4 turned 2 years old last week. That was the beginning of all of this. A door opened and I stepped through. I was excited and anxious. I still am. After two years of trying to learn all I can about AI and how I might use it, I figured what the hell, why not throttle up and put my three decades of storytelling experience to maximum use. 

I realized something a few weeks ago: I have been building this company for 30 years and just didn't know it. I spent thirty years living an artist's life, and the past two years trying to understand AI and how it can help me, and how I can help others.   

Late last summer everything started to come together. It took a few months to tie up some loose ends so I could time things right. The Holidays were my first window. Too much going on to be ideal. Same with January. February may be the shortest month but it was my first clear opening and I've been accelerating ever since. Sure, I've been vibe coding too, but let's not get ahead of ourselves. 

I spent the last week in Savannah. Went to my first St. Patrick's Day Parade. We rode our bikes downtown, got there before noon, and were gone before the drunken shenanigans took over River Street. Who knew there were so many Former Grand Marshalls?

Before my trip, I had been focused on the first of three apps I hope to release before the end of the year. Originally, I thought the third app might have to wait until 2026, but the tools to help create apps have become, well, amazing. Vibe Coding has become a thing and I have been doing my fair share over the past month. That said, I am not trying to spawn a thousand different apps at breakneck speed. I have my plan and will adjust it as necessary. No need to get carried away with every single AI advancement. 

I know Twitter is not what it used to be, and loads of people have abandoned it like a plague ship, but it is still the best source for AI news. And if I am to stay adaptable and in touch with all that is going on I need to be in the know. Because I have been tapped in long enough to some great AI sources (content creators), over the past six weeks I have eased up on scrolling Twitter and have been receiving some of my updates via newsletters or YouTube videos. It's nice to get some context and see some use cases that you don't always get with a company's Tweets.  

Last week, before I left for Savannah, I started to think about more than just the app. I allowed myself the first real chunk of time to think about the company at large, of which the apps are only a part. Someone had asked me back in February if I had come up with a name for the company. I was not worried about it at the time as names or titles for stories kind of emerge. Last week I felt compelled to begin a name quest. Yes, I used quest for dramatic effect. 

In truth, I was winding down from weeks of focus on the app. March is the month I begin to focus on both the app and the creation of the business. Last Tuesday, I started to do some research on names and I began to realize how many of the names I liked were already companies. Many of them are common names that I was not surprised had domains registered. But as I started to dig deeper and deeper I realized that there are a shit ton of companies and all of those that I was looking up, whether they were American or international companies, the names had been used in one way or another. I wanted a name that would represent what I am trying to do: To increase my creative output and to assist others with the knowledge I have gained in this life by utilizing what I have learned about AI over the past two years. The search was frustrating, seeing a name taken one after another. And then I found it. I had to get creative, but I found it. 

We are nowhere near launching, and I've yet to file all the paperwork, but I found a company name that properly represents what I am trying to do and found no mention of the name in use anywhere. Again, I had to get creative, but I had a clear shot at it. So, I thought about the name for a few days to make sure I liked it enough to have it represent both the apps as well as my creative content. Then I took my shot and secured the .com and .net domains. That is only step one of the process, but I found the name of my company, or better yet, it found me. I have steps I still need to take before I can reveal the name publicly, but I will soon enough. Possibly in next month's blog post. 

Part of my settling on the name meant I had to see how it would look. So, I started creating images and logos and began sharing them with a few people whose opinions I value. The name seems to have landed well, and those whom I have shown the images and logos to have given me feedback. Now I have a few workable images and logos. Not only for the company but also for the 1st app. 

I have settled on a few early styles that I like for the Company name, the app name, and the app. My thoughts on the app originally were to make it similar to many of the other tools out there, which are very clean in their appearance. I love simplicity, but I also like style. Do I really want my first app to look like everyone else's app? Yes, but no. While I like the simple appearance of many of these apps, they all look the same. My solution is to keep it simple but with a bit of style. I will probably also have a toggle that will allow users to switch to the more lackluster version. Or not. Maybe I'll just do a light and dark toggle. We'll see. Still some work to do. 

One of my guidelines is to not overcomplicate things. "Keep it simple, Stupid," is definitely a motto I am trying to live by these days. Examples: 1) I created an eight-month and a three-year plan last November. When I realized that the Holidays or January would be terrible times to begin this madness, I moved my start time to mid-February. I adjusted the plans accordingly. 2) Since I made those initial plans back in November, AI has kept on accelerating and making advancements. The writing has been on the wall for months that 2025 would be the year of AI Agents. What was unexpected was how quickly other AI companies would learn from one another to create similar tools.  

That started when OpenAI's o1 dropped in December, a few months after the weaker o1 preview had been released, and introduced us to Reasoning LLM models. We are now awash in reasoning models as other companies quickly figured out how o1 works. OpenAI then dropped Agentic Tasks in January and then Deep Research in February. Tasks are a basic agentic tool but Deep Research was the first proper agentic tool that allows users to create research papers on a subject with more detail than with normal queries. Other companies quickly began following with their own versions.

Those were just the ones that dropped before I began in earnest to create the first app and the company. Ever since the second week in February, the updates have kept coming as far as agents and all the other AI tools. The cost of AI has skyrocketed as well. Not only has OpenAI begun charging $200 for a Pro Tier, but the best AI Video models (Google Veo 2 and Luma Ray 2) are charging an arm and a leg for normal generations. Granted Veo2 is available in YouTube shorts but it is a wonky way to access it. 

The hope is that the cost will drop so that we mere mortals might be able to use these tools to create without having to pay out the ear for the privilege of using them. These tools are there or thereabout where I need them to be to create one of the many shorts I have been thinking about for the past year. My hope is that the cost drops as the tools continue to improve. If you listen to the rumblings, it sounds like Runway and Midjourney may have Video Model updates within the next few months. 

My hope is that they take us over the threshold from good enough for short films to good enough for feature films, TV, and games. What does that even mean? It means that not only do the outputs need to look natural but they need to sound right as well. You can still tell every lifelike AI movie was created with AI. Until that happens it can only be used for VFX shots. Apparently, House of David on Amazon Prime used AI for VFX shots, and that is a pretty big deal. They did not admit this until after it shot up to #2 on their Movies list. That means they were afraid to mention it before its release but willing to give credit once they were a success. The same could be said for Tom Hanks and Robin Wright's Here. Only after its release did they publicly give full credit to the AI teams that helped de-age them. There is a desire by filmmakers to use the technology to tell their stories but a fear of how the public will react. 

Sora is a good AI video tool but it is not what I had hoped it would be. In truth, none of them is as intuitive as I want. I tried out Veo 2 and Ray 2 over the past month when they were made available on one subscription. They were both so expensive that I could not make more than one video each with both models. And I was disappointed with the output video from both models. TBH, it put me in a mood. I knew then that now was not the time to even think about trying to make anything more than a short. It wasn't cost-effective and the tools, while they had gotten better, were not good enough... yet. 

This is the lone video I was able to make with Veo 2 on the paid site I used. If I had created one more I would have burned through half of my credits. This is supposed to be the first shot of the TV series that I tried pimping out last year. Other than the surgical mask dangling before the camera instead of sitting in the middle of the road, not bad. It's not usable but I like the look of the town and the fact that the streets are empty. This is an example of how these models need to get better—smarter. The prompt was not perfect but a smart model should know that the mask is in the street and not dangling before the camera. 

Prompt: A small-town main street at night, illuminated by soft, warm streetlights casting gentle shadows on the pavement. The shop windows glow invitingly, showcasing colorful displays, yet the stores remain eerily empty, hinting at an unsettling stillness. The atmosphere is quiet and serene, with a palpable sense of anticipation for the dawn. In the center of the deserted road, a light blue surgical mask flutters lightly, caught in a faint breeze, its movement subtle yet poignant against the backdrop of the tranquil night. The camera pans slowly down the street, capturing the flickering light reflections and the delicate rustle of the mask, creating a seamless loop that emphasizes the haunting beauty of the scene.

AI video tools are good enough to make an animated movie, but lifelike is still just beyond the horizon. Which is fine by me, as I have a lot of other things to focus on. I do expect I will be able to make an animated short before the end of the year and possibly a lifelike AI trailer for the TV show I was banging on about last summer. The tools were not good enough then but they are now (for shorts), except for the price. Eventually, I'll have to suck it up for a few months and pay the exorbitant fees to get the shots I need. 

My hope is that by this time next month, I will be annoying you to try my new app. If that is the case, then I will begin work on the second app and start work on one of the animated shorts on my list. The one I will start with is the revisualization of a short film I made years ago. That is meant to be the project that helps me adapt to the medium. The other is a story that I had long been thinking about and late last year I finally got around to writing it as a script. 

I am on track with my updated 8-month plan and the 3-year plan. With the help of AI agents and other AI updates, like coding and AI video, I may be able to release the second and third apps sooner than expected. While I have tried Deep Research and Tasks, I have yet to dive too deep into Agents. However, tools like Manus look promising. 

Should there be another big leap beyond Veo 2 in AI video that allows for seamless lifelike outputs with perfectly synced voice before the fall, I may be able to start to work not only on the animated projects and the trailer for the TV series mentioned above but that may also allow me to begin on a feature, TV series or even a game. Games are a big part of the TV series from last year and we will soon be able to create games to release with our movies and TV shows. The cost may be prohibitive but if the tools are available and I can afford them I will begin using them to create lifelike content and games, but not before we cross that threshold. I don't want any uncanny valley scenes ruining my stories, causing viewers to turn the channel.

In review, I am making progress with the app. I also discovered a name for my company that I think embodies all that I am trying to do. Things are progressing. Fingers crossed I can share more next month. I do not like holding back information. I've had to do that about writing projects and now I am having to do it with the app and the company. We're getting closer and I can't wait to tell you more. 

Thanks for reading.

Early Days of Autumn

It has been six weeks since my last blog post. The season has changed and the temperature is starting to cool as we head towards the end of 2024. A few years ago, I was only posting a couple blogs a year because life was just ticking by. I was releasing a new book every few years and working on screenplays and such but it had all become pretty routine. Over the past year and a half things have become far more interesting, and so I found myself wanting to communicate more about it all.  

Six weeks ago, I was wrapping up my review of AI video models. Since then AI Video has taken further steps forward. Runway, Luma, and Kling have improved. Models like Minimax have also been released. I am finally seeing some emotion in the generations. I wouldn't call it acting as the emotions seem more like commands than actual human emotions in response to anything. But it is the first real step towards AI acting out a role. 

I haven't done much with the trailer for the TV series over the past six weeks, mainly test shots with the new features rather than trying to piece things together into something resembling the script I wrote for the trailer. And of course, keeping up with all the updates. My focus has largely been on the APP I have been thinking about for a year and a half. 

To my delight tools like Replit and Cursor have made vast improvements over the past six weeks, which has done nothing but validate my decision to focus on creating my first APP. When I made that decision to focus on the APP as opposed to using GenAI tools to help me create a trailer, or illustrate another book, I had no idea that it would get so much easier for people like myself (someone who has only taken one basic coding class) to actually create an APP. But then that is why I decided to pursue AI with such vigor because it was evident that AI would help in ways that I had thought of and ways I had not, and that the change would continue for the foreseeable future. 

My process remains the same despite these great advancements in APP creation. It will just make it easier for me and may limit how much outside help I will need along the way. I won't be talking about the actual APP until it is fully ready to go but I am more than happy to talk about my process. 

One week after my last blog I saw a video with a young child using Cursor to create an APP. That's when I knew for sure the APP was the best thing to focus on. That was also the same time I had some real-life stuff take precedence for two weeks limiting what I could do regarding R&D with the APP. While I got some planning work done during those two weeks the hard yards were delayed as I dealt with some things. 

I had decided in mid-August that I would use ChatGPT to create a GPT as the first test phase for the APP. During that final week in August, I took another class on Coursera: OpenAI GPTs: Creating Your Own Custom AI Assistants. Once I had completed the course and September rolled around, I was ready to focus on the APP in earnest. 

While I have created several GPTs over the past ten months and thought I had a good grasp on them, I wanted to make sure I was getting the most out of them. The plan was to create the GPT, test it, and then transition to taking what I had learned and separately creating the APP. 

Once I realized that I liked the results I was getting from the GPT, I began testing it and getting some feedback from others. It was behaving how I had hoped, for the most part, and others were seeing the value it provided and appreciated my plans for the APP. 

Because I liked what I was seeing from the GPT I began to wonder if I could use the GPT in the APP or if I had to start from scratch and independently build the functionality I was getting from the GPT. Turns out I can use the GPT in the APP and that my APP will be what is called an AI Wrapper, something I didn't really understand until recently

I asked o1: What is an AI Wrapper? 

o1: "An AI Wrapper is a simple software layer that makes it easier to use an AI model by providing a straightforward way to interact with it. It acts like a protective cover that simplifies complex AI functions so developers can use them without dealing with the intricate details."

Until earlier this month, I had thought of AI Wrappers in negative terms because I believed they were just chatbots masquerading as APPs. Then I began to do some research. I have come to realize that there are a number of really good AI wrappers out there. Some of which I use with regularity but hadn't realized that they were AI Wrappers. The more you know. 

While the APP will likely contain some version of the GPT I have been building and will continue to update, there will be other AI features built into the APP. It is geared to help people and will have Free and Paid features. Hopefully, the new Advanced Voice Mode will be a part of the GPT. As of today it is not. It's only been a day since its release and I can't see releasing the APP for a while so there is a chance.

I don't know what I will charge for the paid services, but my goal is to be able to help people and hopefully make a little money in the process without overcharging people. I have no idea if people will even use the APP, let alone pay for the paid features. 

This whole thing may fall flat and no one ends up using the APP. I would take some positives from the experience either way. I will have learned how to create and release an APP and I fully expect to make others after I finish this one. Another big plus is that I am already using the GPT myself and benefiting from the results. Having done some early tests it has already helped me do something that I have wanted to do for over a year. So, even if I were to abandon the whole thing right now I have already gained something from the experience. But I don't plan on abandoning the APP any time soon. I can see how it can help people and it is now up to me to finish it.  

My goal is to have this APP ready for the public after the election. There is too much chaos to try and wade through right now and it's not anywhere near ready for deployment. I'd like to have it ready for the holidays when people are gathering with their families. I think that might be a good time for release. We'll see if I can actually deliver it by then or if it gets postponed until after the new year. 

Beyond the creation and release of the APP, I am still scouring the internet to keep up with AI advancements. While the APP means a lot to me other aspects of the multimedia company I am trying to create mean just as much if not more. I am a storyteller after all and I want to share the stories I have already written and those I have yet to put on a page. 

Having finally seen some emotion in AI video generation over the past few weeks confirms that progress is still being made and that it will continue. Last week AI Video company Runway signed an agreement with Lionsgate to train a model on their IP. That's huge and won't be the last production company to do such a thing. Mixing live action with AI generations is happening now. Some might say it has been happening for years. It is the way of the future despite all the protests. 

One of the greatest film auteurs of our time Guillermo del Toro is not a fan of AI at all. And who can blame him, he's still at the top of his game and has not needed it up until now. While I respect the hell out of Guillermo I cannot make a pledge to continue on as I have for 20 years with little traction and expect things to change for the better. I would just have more scripts sitting on a shelf that would never get read let alone made into movies, especially with fewer and fewer production companies refusing to use AI. That's why when I saw James Cameron was joining the board of Stability AI, I felt there was a balance to be had with GenAI when it comes to filmmaking. One that will likely continue to shift over the next year or so. There will no doubt be companies that will refrain from using AI, but that number will continue to dwindle. 

The window to the past is narrowing but the one to the future is expanding by the day. Nothing would have brought me more pleasure than to have had the chance to work with someone like Guillermo on one of my stories, but that just hasn't happened. Yet I still have a desire to tell my stories and if you think I'm going to play the Hollywood waiting game anymore when I don't have to then you are fuckin' crazy. I need to see my stories come to life and I won't let the current opinion of one of my idols keep me from pursuing my own dreams. 

Yes, I will keep working on the APP, but creating Movies and TV shows with the help of AI tools is getting closer and closer every day. As of six weeks ago, I thought that you could only make a believable AI movie in animated format by the end of the year. Now, with these recent advancements, we are getting closer and closer to AI live action movies. We're not there yet but things may still advance enough in the last three months of the year that we may cross that line of believability that the general public will accept as real. There likely won't be enough time to make that movie by the end of the year but the tech may be advanced enough by then for the production of that first believable AI movie to begin production. Maybe. 

If last year was anything to go by, the release of GenAI improvements will slow the closer we get to the end of the year and may not pick back up until February or March. So, if we don't see some cool new upgrades by early December we likely won't see them until a couple months into the new year. But what do I know? I'm just trying to keep up with it all so I can continue to course correct when it is appropriate. 

And with the help of Cursor and Replit creating and deploying an APP has never been easier, which is why that has become my main focus for the time being. I owe it to myself to stay the course with the APP until its release, despite any AI video upgrades between now and then. Thanks for reading.