[flipdin]

I import physical products from China and sell them online in Australia. Would prefer not to disclose my niche, but it’s the equivalent of selling bedding/pillows. Nothing at all unique or interesting, the products are good and well priced though.

[digikata]

Are you running the direct data of your workload all through the AI? The number of tokens you're consuming can probably be reduced a lot. It seems like a good time to take a half step back and use the AI to build a workflow engine to shift more of the work out of the AI models and into more traditional scrape and analyze sw logic. For the more sophisticated decisions, even using a LLM grade AI to bootstrap more older machine learning algorithms and rules could AI use to updates of rules instead of something called inline on every working transaction.

[flipdin]

When I had that many subscriptions, I was being very lazy with it.

I do build a lot of automations though that don’t require AI. By that, I mean using AI to create python scripts or dashboards that can do the same task over and over.

It’s just convenient and easy to type stupid stuff like “find my email with Jake that has packing list for order XYZ, print it on my downstairs printer”

I know that’s lazy and way overkill for fable or even opus… I would run stupid prompts similar to that all throughout the day when I didn’t want to get sidetracked from whatever my main focus was.

But you’re right, and I have been handling just fine with 2 x codex subscriptions.

I actually just need to build a router too, that uses cheaper models for my basic requests. My token usage could do with a lot of optimisation.

Also, I was just using 3 accounts, but when Fable was released for a short time, I just figured I’d spam it while it was cheap with a lot of projects opus was struggling with.

[nunobrito]

Get a computer with a good graphics card and try out opencode. The quality has recently improved a lot (comparable to Claude about 12 months ago) and everything is running locally, so you'd save on costs but most important: nobody can kick you out.

[haute_cuisine]

Author wrote that opus required too much handholding, to host opus locally (aka glm 5.2) you need $50k in hardware (or $3k/mo to rent a slow server) or just buy tokens from hoster.