My AI cut a TikTok clip, wrote the caption, and emailed it to my phone to post

Setting up the Task

I needed to post a TikTok clip for my website's social media promotion, and I decided to put my self-hosted AI system to the test. The task involved cutting a video clip from a longer video, writing a caption, and emailing it to my phone for posting. I have 20+ websites to manage, and automating tasks like this can save me a significant amount of time. My system consists of 50 agents, including the overseer (Atlas), content agents (Axiom and Forge), and SEO agents (Sentinel and Scout), all running on a single RTX 4050 6GB laptop.

Assigning the Task to Agents

I assigned the task to Axiom, my primary content agent, which is responsible for generating text and video content. Axiom has access to my video library and can use its video editing capabilities to cut clips. I also involved Forge, my secondary content agent, which specializes in caption writing and optimization. The overseer, Atlas, was responsible for coordinating the task and ensuring that all agents worked together seamlessly. I used cron jobs to schedule the task, allowing my agents to work in the background while I focused on other tasks.

Video Editing and Caption Writing

Axiom cut the video clip using its video editing module, which is based on a Python library. The clip was 15 seconds long, and Axiom was able to trim it down to the required length in under 2 seconds. Meanwhile, Forge generated a caption for the clip, using its natural language processing capabilities to come up with a catchy and descriptive title. The caption was 150 characters long, which is the optimal length for TikTok captions. I was impressed by the quality of the caption, considering it was generated by an AI agent.

Emailing the Clip and Caption to My Phone

Once Axiom and Forge had completed their tasks, Atlas took over, emailing the clip and caption to my phone using a Python SMTP library. The email contained a link to the clip, which was uploaded to my cloud storage, and the caption in the body of the email. I received the email on my phone and was able to post the clip and caption directly to TikTok. The entire process, from assigning the task to posting on TikTok, took under 5 minutes, which is a significant time savings compared to doing it manually.

Trade-Offs and Limitations

While my self-hosted AI system performed well, there were some trade-offs and limitations. For example, the video editing module used by Axiom is not as powerful as commercial video editing software, and the caption generated by Forge may not be as creative as one written by a human. Additionally, the entire process relies on my agents working together seamlessly, which can be a challenge when dealing with complex tasks. Some of the key limitations and trade-offs include:

• Video editing capabilities: 2 seconds to trim a 15-second clip, but limited to basic editing functions
• Caption quality: 150 characters, optimal for TikTok, but may not be as creative as a human-written caption
• Agent coordination: relies on Atlas to oversee the process, which can be complex and prone to errors
• System resources: uses 10% of my laptop's CPU and 20% of its GPU, which can impact other tasks

Conclusion and Future Plans

In conclusion, my self-hosted AI system was able to cut a TikTok clip, write a caption, and email it to my phone for posting, saving me a significant amount of time. While there were some trade-offs and limitations, the benefits of automation and increased productivity far outweigh the drawbacks. In the future, I plan to continue improving my agents' capabilities, exploring new libraries and modules to enhance their performance. I also plan to expand my system to include more agents and tasks, further automating my workflow and increasing my productivity. With 20+ websites to manage, every minute counts, and my self-hosted AI system is an invaluable tool in my workflow.