Hi! My kids turn 3 and 1 next month and if you have kids in this same age group I would looove your tips on how to consistently make content when there’s always crying or screaming or laughing going on in the background 🥲

One thing I keep noticing lately is how easy it is to overcomplicate this whole process. There are now so many AI tools, automations, editing apps, schedulers, workflows, platforms, prompts, subscriptions, and strategies that it can start feeling like the actual content becomes secondary. But one thing that has stood out to me is how often simplicity kept coming up. Simple image posts. Simple workflows. Simple systems repeated consistently. A few people even mentioned that some of their better-performing content was not their most advanced or time-consuming content. I’ve honestly been reflecting on this a lot lately because I think many of us are trying to build something sustainable while also working full time and managing real life responsibilities. Sometimes simplifying the process may actually help us stay more consistent long term.

I used to put carousels off for weeks at a time. 8 slides, each one needing its own graphic, and by slide 4 I'd lost the will to finish it. Now I don't open a design tool at all, and the reason isn't that the tool got better. It's that I stopped starting from a blank canvas. I start from one of three templates instead, and those three get all the use. The instagram carousel slideshow, a whiteboard infographic that suits linkedin, and a manga panel one I keep for livestream announcements. Then I give it the source, which can be screenshots, a post that already worked, or just a topic. It writes the copy and generates every slide from there. The slides come out of nano banana 2, which matters more than it sounds. if you ask for a carousel and you'll get 8 slides that don't quite match each other. What makes things work for me is i write the prompt for each slide, or describe the visual style once and hold it across all of them, and they come out looking like a set. That's the difference between something you'd post and something you'd quietly delete. You can also drop a slide without rebuilding anything. I generated one where the text on the first slide came out mangled, so I said get that carousel but remove slide 1. The 6 that were fine went out as they were Then the same asset goes out as an instagram carousel, a linkedin carousel, a tiktok slideshow, a pinterest carousel and a facebook carousel. I make these in blotato and schedule them from the same place, so the carousel never gets downloaded and re-uploaded anywhere. canva is genuinely good at this too, if you'd rather lay them out yourself.

I've spent the last few weeks running fable 5 on real jobs across my business and codebases. with the $50/M output tokens price tag, running standard 1-step prompts through it is just a fast way to burn money. If a task takes 10 mins or fits in a simple chat window, opus 5 or sonnet is fine. fable 5 is built for ambiguous, high-context jobs. When I paired it with tool harnesses like claude code and custom subagents, I started seeing where the price tag actually pays for itself. Here are the 5 use cases I tested and what I learned: 1. CLONING PAID SAAS INTO LOCAL APPS Every month I look at my software bills and wonder why I'm paying $30 to $50/mo for single-purpose web tools, esp ones that send my private voice recordings to cloud servers. i then used fable 5 to build a local alternative to tools like wispr flow: - wired up local whisper directly to macos system hotkeys - dictation stays 100% on my laptop with zero network latency - costs nothing per month and keeps data private What would usually take hours of messing around with native macos apis and python audio bindings, fable 5 built and tested in one autonomous run. Now when I see a small SaaS tool, I describe the workflow, hook up airtable + tally, and let fable 5 generate the UI and glue code in one pass. 2. CLEANING AI MEMORY CLUTTER (MEMORY-GC) Over time, my agent's memory .md file bloated into a 15KB junk drawer. Every prompt was getting slower and picking up stale context from weeks ago. I pointed fable 5 at my session logs and memory files and asked it to run a garbage collection pass (memory-gc): - audited the file and stripped out redundant context - turned memory into a clean routing layer instead of a warehouse dump - extracted repeated prompt steps and compiled them into reusable agent skills Now I run those skills daily using cheaper models like sonnet. Using the frontier model to build infrastructure for cheaper models is peak leverage imo. 3. THE OBSOLESCENCE VS 10X STRATEGIC AUDIT

I used to spend most of my day typing at a model, which is prompt, read the answer, then prompt again. I was faster than I'd been without it, and still sitting inside the loop. Now two agents run on my support inbox every morning. The first closes open tickets through the intercom API at 8am. The second reviews everything it closed at 10am, and reopens anything that shouldn't have been. It's doing the job I used to do. Getting there took months, and what blocked me was never the wording of the prompt. WHAT I TRIED FIRST: I appended "don't stop until..." to the end of a long prompt. That works in the sense that the model keeps going. It doesn't work in the sense of telling me whether the result is right, since the thing doing the work is also the thing judging it. This is the whole problem. WHAT ACTUALLY FIXED IT: The fix is a checker inside the loop, not a better prompt. It has 3 parts and each one does a different job. - An end state instead of a task, so there's something concrete to finish against. "Sort every file in my downloads folder into subfolders by type, keep going until no files are left" - Something countable it can check itself against, since a model grading its own prose will always pass itself. Zero loose files, zero rows without a label, tests passing - A guardrail so it can't do damage while nobody is watching. Don't delete anything, stop after 30 turns That's the /goal command in claude code or codex, where a small fast model checks the condition on every pass and then verifies the whole thing at the end. 5 GOALS I'VE ACTUALLY RUN: These are 5 I've run myself, so you can see the shape of them. - Rename every invoice into this format in date order, until no file has its original name. Stop after 25 turns - Categorise this spreadsheet until no row is missing a label - Work through this stack of documents until none are left - Clean up these captions until they all fit my brand voice rules - Turn each of these rough ideas into hooks until the list is done