Chapter 12

A sense of accomplishment washed over Henry as he stared proudly at the final product. Except for eating and sleeping, he'd been working non-stop on this project for days. Even with the knowledge he acquired, he never made any mistakes in his code except a few typos which were caused by his mediocre proficiency in typing yet it took him over a week of writing continuously to complete it.

"Haha, finally complete," Henry said feeling on top of the world. This was something that if asked to do a few months ago, couldn't dream of it. His hands were tired from all the typing and occasionally got finger cramps. Henry didn't regret any of the time spent as he felt it was all worth it.

For the next few hours, Henry was engrossed in testing the AI. Although he knew it worked flawlessly, he still kept on tinkering with it. It was his work and his pride so he savored the moment.

The web application was also completed. It had a simple and minimalistic design and at the mid-bottom was where the prompt was entered. Currently, the standard AI wasn't much different than the other ones on the market but what would make his AI stand out was the Image and video generator.

He integrated Thought Visualization Synthesizer (TVS) into his web application and whenever an image prompt is entered, it goes to Thought Visualization Synthesizer (TVS) and gets generated before sending it back to the user. Whenever a follow-up prompt is entered, it goes to TVS and gets updated there.

The AI itself didn't need to understand or analyze the images since it was entirely done by TVS. The main thing he lacked to make this AI incredible was training data. The training data for image and video generation could be easily fixed, He had already integrated a learning algorithm into the AI, designed to analyze all prompts fed into it along with the image feedback from TVS, using this data to refine and expand the base training model.

The same applies to video generation. While it was essentially still image generation, it required generating 60 frames per second of video. This meant video generation demanded significantly larger training data. Additionally, it needed to analyze minute changes in motion, such as a mouth opening or a finger moving, to ensure consistency and avoid disfigured or unnatural outputs in the final video.

So whilst the AI itself was almost useless, given enough time, the AI would become a very powerful tool. Now he only needed to set the server up. If he were to run everything on his computer, before the userbase reaches 100, his pc would encounter crashes very frequently.

Before Henry could start setting up the server, he made a call. It was Michael whom he met at microcenter. After making the call, Michael picked up a few seconds later and Henry began asking questions to clear his doubts.

After ending the call, Henry felt a bit clearer on some things. It certainly wasn't cheap to host a server. After inquiring from Michael, he recommended going for Amazon Elastic Compute Cloud(EC2) and he should choose whatever hardware requirements fit his project.

Henry selected the following after contemplation:

AWS m6a.16xlarge (64 vCPUs, 256 GB RAM) - $2100

AWS p4d.24xlarge (8 NVIDIA A100 GPUs, 96 vCPUs, 1.1 TB RAM, 8 TB NVMe SSD storage) - $26000

Amazon S3 (Simple Storage Service): 50 Terabyte (TB) - $1050

Amazon RDS (Relational Database Service): AWS Aurora MySQL, 32 vCPUs, 244 GB RAM. - $3500

Elastic Load Balancer (ELB) for distributing traffic across servers - $100

Data Transfer Out (Internet Traffic): 10TB per month - $1000

TOTAL: $33,750

Indeed, hosting a server wasn't cheap in the slightest. He wouldn't pay for it till it was time to go live. For now, he still needed to think of a way to promote it. Naturally, Ads would be the way to go but he'd never seen any situation where an ad led him to use a project so he abandoned that idea.

Another idea he had was to reach out to Giant companies and look for a partner. What he was bringing forward had a lot of potential so many would be willing to invest in it. Most AI can only make short videos and they still have issues sometimes but this AI could enable one to make a movie supposing they know what they're doing.

With that, Henry started composing an email. He'd send an email to different companies and hope to get offers but he was optimistic about getting an offer. He was so confident about receiving an offer that he was already thinking of rejecting offers to buy his AI.

Henry spent a few minutes composing emails to send to each company he was interested in, he sent one to Google, Microsoft, and IBM. He didn't plan to send it to many companies. He left the mail in draft since he wanted to do something first. 

Henry also got the domain ready for his website along with a custom email for Phantma. He wanted to look professional at least.

Although he won't buy the main servers yet, he'd still get a small-scale server that'd last a month and use it to host the AI. If they decided to check out his AI to see if it's worth partnering, they'd be able to do it for a month. After that, if he didn't get any responses yet, he would find different means to grow.

He purchased a small server with specs similar to his pc for a month which cost $500. With that done, he gave the link to his website his AI application is based. It was running on a GitHub domain temporarily but when he's fully started, he'd move everything to a more stable environment.

With everything done, he sent an email to each of the companies.