AI project - Jan 2026

I've decided to spend more time this year on AI projects. Ideally 1 each month but we'll see how it goes.

For Jan 2026, I've decided to focus first on social media content creation specifically my podcast and building a faceless automated channel.

Key Insights -

1. It still takes some time and effort to build out an automation if you come from a non-technical background. I thought it would be faster but the actual setup takes up time. So 'AI' advantages, comparative advantages due to this new technology, will last a little longer than it seems.
2. It's not all laid out in an easy way. Powerful tools now exist but to get software to talk to each other is hard for someone coming from a non-tech background. Getting applications to talk is key.
3. Time, Action, Energy increase in value. Knowledge, Money, Credentials are slightly less valuable (but still valuable). One of the major constraints right now is to actually be able to take out time, to do something and not get burnt out. AI is actually beneficial for younger people due to this.
4. AI is a tool that is to be built into workflows. It improves the process clearly but what should be the goal of the process is still not an easy AI win. I can build a faceless content channel but why should the content be about? AI gave me ideas but it wasn't very out of the box, creative or something that might succeed. It's building from the past. Due to this, at least in the short term, creativity, out of the box thinking, gets more valuable.
5. Sometimes a bottleneck can be solved by another AI. I was finding it difficult to render videos after the content was created into one video. Popular bots suggested paid ways to get software which did not make economical sense. While Poe's assistant suggested ffmpeg which worked out great.
6. Claude Code is really good. It's having a personal software engineer work for you who is super smart. It's very beautiful to see it running and building out. It does use a lot of tokens and in a very short period of time usage can shoot up.
7. Poe is another great tool. Via its API you can connect to over 300 models and have a centralized billing. It also offers some discount on API costs for some models.

I succeed in creating an automated channel and saved some time that goes into my podcast. Here is the channel if you're interested. It contributes to AI Slop and unless it succeeds in being interesting I will shut it down in about 2 months. The focus was more on building it via AI and automation and less about the content. The content I admit is not that good and I did neglect it.

So the key now is focusing on what you're solving for, what people are interested in. If you find something that does fit, the building out is easier, faster and cheaper than ever before.

I am have so many ideas to execute but feel a constraint on time and energy. Nevertheless, I am super excited about this and just want to spend every waking minute working on these projects.

Here is the automated channel and here's my podcast