Activity
Mon
Wed
Fri
Sun
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
What is this?
Less
More
31 contributions to AI Automation Society
Order of Operations (Claude Code)
Hi all, After a conversation yesterday, I developed an "Order of Operations" flow map that I added to LinkedIn. I know a lot of people struggle with AI (Claude Code) doing what they want and, over time, experience more variation and AI hallucinations. Often, we contribute to the problem because we don't understand its processes; we don't understand what it needs from us. Go there, download the detailed flow and connect with me. LinkedIn Post: https://www.linkedin.com/posts/johndmackenzie_claudecode-anthropic-aiengineering-share-7493232665089818625-kFDC/?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAmh2roBgwSjoTy-2kCbi0MHaOh-BW9HYgE
Order of Operations (Claude Code)
1 like • Aug 12
@E H interesting. Other things that impact is session length, crossover memory, as the more you use Claude code the greater the chance of conflicting statements, thus the AI needs to make decisions it will often get incorrect… which leads to AI Slop and Hallucinations… at least from a human perspective.
1 like • Aug 12
@Marwin Emia i have similar processes
Is it possible to make long Youtube vid completely with claude?
I’ve watched a few video tutorials, both from Nate and other creators, but I still don’t quite understand the process since they only demonstrate how to create 1–2 minute videos. If I want to make a 30 minute video, do I need to generate each part separately and then combine them? Are there any tutorials I can refer to? Thanks!
2 likes • Jun 26
I'd be careful with generating a full 30-minute video entirely with Claude in one go. The common approach for longer content is indeed to break it down into segments, generate each part, and then assemble them. A practical next step would be to experiment with generating a 3-5 minute segment first and see how Claude handles the continuity and detail before attempting the full length.
🚀New Video: Claude Mythos is Finally Here.
Anthropic just dropped Claude Fable-5, a Mythos-class model that sits above Opus and is available to everyone right now, plus Claude Mythos Five for Project Glasswing partners only. I break down what the two models actually are, the pricing, why Fable-5 is only free on your plan until June 22nd, and the benchmark jumps in coding, knowledge work, and vision. I also revisit my earlier Mythos prediction and what I'm most excited to put Fable-5 to work on.
1 like • Jun 9
These model providers are blatantly taking the profiteering. No added cost, we'll take it away and then we'll bring it back... Double, triple... 5x prices... Claude has many people moving elsewhere and using other platforms, and then only using it for specific work, which is undermining their own revenue.
Most people overcomplicate prompting. The real leverage is simpler.
After working a lot with LLMs, I’ve noticed something consistent: Better outputs rarely come from “better wording”; they come from better structure and intent clarity. A few things that reliably improved results for me: - Start with clear context and a concrete goal, not just a role - Put relevant background first, question last, especially for long inputs - Make the task single-purpose (one prompt = one outcome) - Ask the model to clarify before answering if anything is missing - Define what “good” looks like upfront (constraints, tone, output shape) - Let it do a quick thinking pass before the final answer (approach first, then execution) One shift that made the biggest difference for me: Treating prompting less like "writing instructions" and more like designing the interaction flow. You don’t need perfect prompts, you need fewer ambiguities. Curious what’s been the biggest improvement in your own prompting?
1 like • Jun 8
This is a great breakdown! I'd add that the "define what 'good' looks like" point is particularly key. A simple test is to give the AI two definitions of "good" for the same task and see how its output changes.
HIRING: Back-end Automation Developer
We're running an AI-software development company with currently 9 developers. We're looking to expand with talented back-end automation capacity. Mostly back-end automations orchestrated in n8n, while leveraging custom coded python scripts, calls etc (AWS), and integrating with third-party tools via their API docs (not always very clean) and of course recreating company logic (with AI + business rules). Some full-stack experience is a pre, but MOSTLY back-end automation engineer, coding experience is a Must. High-quality English is a must, availability soon. Check the link below for more details and apply: https://tally.so/r/b5q1y2
0 likes • Jun 7
[attachment]
1-10 of 31
John Mackenzie
5
320 points to level up
@john-mackenzie-2183
Trying to figure it out. Business & Cyber dude Ideas, Methods, Research, Books, Apps, Solving Problems https://jmactech.com/ https://jmaclearning.com/

Active 3d ago
Joined May 28, 2026
Powered by