Back to blogYouTube Video

Published August 15, 2026

your automations suck because they lack granularity - Automation Tip #2

Can't play the video or having issues? Here's the direct link.

AI Summary

The core argument is that automation reliability is directly proportional to the granularity of the process definition. Attempting to automate a workflow without a microscopic understanding of every discrete step inevitably leads to missed logic, edge cases, and brittle systems. The creator illustrates this through the evolutionary development of their video editing pipeline: it began as isolated scripts for audio normalization and noise reduction, expanded to AI-based silence trimming (Silero VAD, Whisper), and later incorporated resolution downscaling—each step added only after manual repetition revealed the hidden complexity. The ability to articulate the entire chain end-to-end is the prerequisite for successful automation; if you cannot describe the process in extreme detail, you do not understand it well enough to automate it.

Key Takeaways

  • Granularity is a proxy for understanding: The ability to describe a process in minute detail proves you comprehend the full dependency chain; inability to do so signals missing steps or hidden complexity.
  • Muscle memory masks complexity: When performing a task manually, the brain automates micro-steps (subconscious competence), making them invisible to conscious analysis. You must deliberately deconstruct these 'automatic' actions to capture the true workflow.
  • Build automation incrementally, not monolithically: Start by automating the smallest, most atomic unit (e.g., audio normalization). Validate it, then chain the next atomic unit (e.g., silence removal), progressively assembling the pipeline as understanding deepens.
  • Manual repetition is a discovery tool: Do the task manually multiple times specifically to uncover the 'in-between' steps that are easy to overlook but critical for logic flow.
  • The 'Explainability Test': Before writing automation code, verify you can walk through the entire process step-by-step—inputs, transformations, decision points, outputs—without gaps. If you cannot explain it, you are not ready to automate it.

Description

Book a call: https://calendly.com/itshassanaziz/discuss-a-project ==== ==== ==== Hey all, starting a new mini-series on my channel. Automation Tips. I've been building a lot of automations recently, both for myself and for other business owners. Unlike 99% of automations you see going viral on Twitter and IG that barely work beyond a demo, my automations are fast, reliable, and cheap. They don't cost $50k in LLM tokens because I know how to write efficient automations. They run fast. They have maximum deterministic capabilities so they're as reliable as possible. I've learned a lot after building these automations, and I'm gonna be sharing all of that in these mini tips.\n\nHope that helps y'all in your own automation projects. Make sure you SUBSCRIBE so you don't miss any part of this series, and leave a like (it helps a ton! Especially for small channels like mine lmao) ==== ==== ==== LINKS Website: https://www.hassandev.me Portfolio: https://www.hassandev.me/work YouTube: https://www.youtube.com/@itshassanaziz?sub_confirmation=1 My Book: https://www.hassandev.me/designing-websites X / Twitter: https://x.com/intent/user?screen_name=nothassanaziz

Transcript

Auto-generated transcript
all right guys automation tip number two now this is a series that i'm doing on my channel where i'm just going to explain a bunch of short and simple automation tips you can apply instantly to your ai workflows and agents and they're going to make them a heck of a lot more reliable and so with that out of the way the second tip that we're going to discuss today is that more granularity is good the more detail in which you can describe your entire process you know the process that you're trying to automate the better it's going to be and the more reliable it's going to be and the easier it's going to be to automate it. Because look, when you can describe the entire process in way more detail, it means you have a much more intricate understanding of all the steps that go into this process, right? You understand every single tiny step in the entire chain way more deeply. Not only that, but it also means you have a much lower chance of missing some critical step in the process, right? And you'd be surprised how common that is, right? If you can't explain the entire process in a very deep level of detail, you're probably missing a couple of steps, right? If you're doing this process manually right now, you're probably just getting used to it, you're developing the muscle memory, and you're forgetting all the tiny things you're doing in between, right? So if you can describe that process in a more granular level, you're going to have a much easier time automating it. Otherwise, you're just going to miss critical details and steps, and it's going to become a problem later, right? And the best example for this is, like, the best example I can give you about this is just my video editor over here, which is just automatically editing and posting all of my content for me, right? This has some pretty advanced automations, but it didn't start out that way. I started by having really tiny automations that automated some tiny task for me and then I just chained them together and kept adding more as I understood the process better right Like I was like at first I just need to process the audio right So I added a bunch of scripts to process the audio to normalize it to apply noise reduction to do all sorts of other things, right? And to then attach it back to the original video, right? That's one tiny step of the entire pipeline. Then I figured out how to trim the silences using AI models like Cilero, VAD, and Whisper, and all these other things, right? And so I added a step for that. Then when I invested in better cameras and equipment, that i got actual 4k videos over here i had to downscale them to 1080p so i could edit them and so i built a step for that right none of this happened all at once right i had to figure out each step in the entire chain i did it manually multiple times so i understood the entire process way more like in way more detail and then i was able to automate it and if you if you were to ask me exactly what's going on in this entire code base i could explain it to you in such deep detail because I understand the entire process so deeply like I just take you step by step in the entire pipeline through the entire chain exactly what's going on how a video is fed into this chain and what the what each step is that it goes through and how it outputs a final video file for us I understand the process very deeply and that is exactly why I'm able to automate such complex things and that's my tip for today always try to describe the task you're trying to automate in as much detail as you possibly can. Because here's the thing, if you can't describe the task in really, really granular levels of detail, you probably don't understand it well enough anyway. And if you don't understand it, you're not going to be able to automate it. Simple as that. So always make sure that you can describe the task in very precise details. Once you can do that, and once you've proven that you actually understand the entire process, go ahead and automate it. At that point, it should be very easy for you.

Share this article

All great things started with a conversation

If you've got a cool project or opportunity and you want me to be a part of it, set up a free meeting with me here, and let's talk. 😊