making local llm-infused mac apps is really fun
making local llm-infused mac apps is really fun
any todo app / work system recommendations?
tried everything but with context switching between multiple private projects / 9-5 / private life etc I always end up in a mess ๐
Claude Cowork is underrated. It thinks more clearly and delivers better solutions.
pov: building a product in Germany
all my side hustle profit is gone for tax account invoice
roughly 2k every year ๐ฅฒ
i tried doing it with claude cowork, and it gave me the exact same numbers as the tax consultant, already in an excel sheet, ready to enter and send to the german tax office
here is the video, highly recommend it: www.youtube.com/watch?v=SaWS...
Better late than sorry.
If you don't have a good guide or tutorial, setup could take some time. Thankfully, people in @tinkererclub shared the perfect video.
have some ideas I want to try out
funny how this works
started using manus heavily a few weeks ago for research and automation tasks, really great execution layer
now Meta acquired them for $2B
hoping they keep it as good as it is
been testing a bunch of ai image generation tools lately
some are great for certain things, others just miss the mark completely
what's your go-to right now?
looking for something with exceptional ux, not just good outputs
watched the traffic patterns from my last pSEO attempt and this time I'm actually prepared for the scale
let's see if it catches any of that search volume
four months ago I was waiting forever for pSEO to work. now I can spin up 50 new landing pages in an afternoon
best part? I can add seasonal keywords before they peak. christmas stuff in november, valentine's in january, you know the drill
the migration, the CLI scripts, the image generation pipeline, translation system for 11 languages... all working together now
you add keywords via CLI, it hits the AI Media Studio API, generates images for each keyword/subject combo, then auto-creates pages at /wallpapers/christmas, /backgrounds/merry-christmas, etc
each page gets proper metadata, structured data, sitemap entries, the whole SEO package
finally got my pSEO system working the way I wanted
built a whole keyword system that auto-generates landing pages for high-volume search terms like "merry christmas wallpaper" or "happy birthday images"
the setup is pretty straightforward but took me some hours to get right
sometimes you just have to accept the incompatibility, respect the difference, and drink your coffee.
how do you handle such situations?
he dismissed all of it. for him, it's corrupted or manipulated.
I get it, yeah, false info has been spread in the past, science isn't perfect.
but evolution theory seems pretty solid to me ๐
stopped trying to argue after that.
for me, truth comes from data, research, iteration.
for him, truth is static and pre-written in scripture.
he said something like "every religion mentions the afterlife, so it has to be real."
I brought up research, evolution theory, scientific documentation.
you can't use logic to dismantle something that wasn't built with logic
learned this the hard way yesterday with an old friend
he's a creationist. believes 100% in adam and eve, rejects evolution completely, expects judgment day.
tried to debate it, but hit a wall immediately.
decided to let the data make the call. spinning up pSEO pages + @outrank_so blogs to see if I can replicate the traffic patterns from ai-media-studio.
giving it a few months of content autopilot to validate the demand before I fully commit to the rebrand.
new landing page for @ecomdesignlab is live powered by @Lovable_so
currently fighting the urge to niche down vs go wide.
"AI Fashion Toolkit" is safe, but "AI Ecom Toolkit" feels like the actual goal. fashion would just be one vertical.
curious tho, how does AI coding perform with tanstack or other frameworks? anyone notice a difference?
don't really care about framework wars but next.js is just stupid easy for me
backend and frontend in one place, tons of templates, and because it's so popular and open source, AI coding assistants know exactly what to do
claude code and cursor just work with it
@polar_sh I know you also have Usage-Based Billing and Credits. My logic was based on the time before it was released. How do you manage these edge cases? Could I solve it in the same way?
this took way longer than I thought but now it just works. users can upgrade/downgrade freely without worrying about losing money or credits.
building billing systems is not fun but getting it right feels good
credits were another rabbit hole. upgrades add the difference immediately. downgrades don't touch credits until the actual downgrade happens. renewals give full plan credits.
Polar handles the actual subscription changes, Supabase tracks the status, QStash handles the delayed execution.
stored everything in a table with status tracking: scheduled, executing, completed, cancelled, failed, skipped
also had to handle edge cases like what if subscription already changed by the time the job executes? or what if they canceled their sub entirely?
the tricky part: what if user changes their mind? had to build cancellation logic. if they upgrade before the scheduled downgrade executes, cancel the scheduled job. if they manually cancel it, delete the schedule queue.
so downgrades are scheduled. user clicks downgrade, system schedules it for end of current billing period. they keep all their credits and benefits until the period ends.
initially thought: just change the plan immediately like upgrades. but then realized that's terrible UX. user paid for the full month on the higher plan, why would I downgrade them right away?
spent way too much time overthinking subscription upgrades and downgrades for ai-media-studio
here's what I learned building it:
upgrades are simple. user clicks upgrade, charge them prorated amount, add credit difference immediately. done.
downgrades broke my brain for the last hours.
Find an area such as SEO, Pinterest, or TikTok, and become skilled at gaining views and traffic.
the truth nobody wants to hear: your dev skills get you to 20%. marketing gets you the other 80%.
think about your marketing plan and how you'll reach people before you write a single line of code
the best product with no one knowing about it loses to the okay product that everyone sees