Zero tech background, building with AI.
Juggling shifts, a travel business that sounds way more glamorous than it is, and a 3yo.
Built 3 apps. Made mistakes on every one, but each build gets sharper.
Now testing build 4: a viral game experiment.
Goal: Β£0 to Β£5k MRR β
In my travel business, I forced myself to do in-person networking events and hated every second. Small talk and pitching myself in a crowded room is just not me.
On X, it's completely different.
You share an honest thought, it resonates with someone 3,000 miles away, and you end up swapping build feedback in DMs.
Sent my first DM to another builder yesterday just to help with their launch.
Have you found online founder communities easier than real-life networking?
When people build gamified apps, they usually focus on points and leaderboards.
The real hook is tribal psychology.
People will fight 10x harder to defend something they have in common with a group than they ever will for generic points.
Competition hits different when there's collective pride on the line.
What's a mechanic in an app or game that instantly hooked your competitive side?
@laozhang_ai finer details and nuances. It already had an exact spec to copy that had been produced in Astra Low . But this was like Picasso vs a 3 yo child.
My usage just went from 100% to 30% of the 5hr limit on Astra Light without it doing anything.
Like I know it burns token faster but that is just crazy.
Definitely some sort of error surely.
Time to use one of those weekly resets i think .
@heypetar I'm still building πNothing to ship yet. This will definitely be a V1 with features and improvements added later.
I'm being more strict on this build that others and sticking to the original scope. It's painfully close to being ready to put out into the world π
@jp54362 I got about 10 minutes before hitting 5hr limit on Astra Medium. Dropped down to Astra and Sol xhigh . Didn't think to increase the model . So backwards
This seems crazy.
Who would have thought to up the model and not lower it to save tokens ?
Makes no sense.
Anyone else tried this.
Thinking about testing it out but don't want to be burnt
π¨ Confirmed, GPT-Astra MEDIUM IS SOMEHOW MORE EXPENSIVE THAN xHIGH
I tested both reasoning levels myself
Medium burns through usage insanely fast xHigh uses noticeably less
Higher reasoning is somehow MORE efficient
Low/Medium were supposed to save usage But xHigh might actually give you more mileage
anyone else tested this?
@bytegrad I've just seen that Astra xHigh is apparently way more token effiecient than Astra Medium. Seems backwards and i've not tried it myself in xHigh but lots of builders are saying it . might be worth a test?
@buildwithrajath This makes no sense. I used Astra Medium obviously thinking it would use less tokens that xHigh. Medium burned through tokens so i didnt think to test xhigh. It seems backwards
@NicolasZu I used Astra medium and it lasted about 10 minutes it was crazy. Didn't even think to attempt upping in to xhigh as i assumed that would be worse. Might give it a test but seems backwards
@heypetar Yeah life's quite full on atm.
My current build is a gamified experiment tapping into tribal psychology and team competition. dropping the first look in a couple of days.
Starting to think about best to launch and get it out there.
@lucagalvani@X this one is something completely different to my usual builds! it is a gamified experiment tapping into tribal psychology and team competition. dropping the first look in a couple of days.
what kind of stuff are you working on at the moment?
@laozhang_ai I managed to get some comparable results on Sol Extra high.
I then asked for a more complex design that I thought it should be able to handle but it was way off. So I switched to Astra just to polish.
Managed to get alot more done by switching the models
@dawbuildsthings Just checked your build out. Great concept.
I always struggle with boiling things down to just a keyword this solves a real problem.
And the report you get plus the tailored emails ππ½.
Will defo have a little try of it
Building the app is the fun part. You get into flow, solve logic problems, and watch it come alive.
Then it's finished, and the cold reality hits: where are the people who actually need this?
Finding users is 10x harder than building the product.
I would pay serious money for a tool that genuinely understood my product, voice, and audience, and handled distribution on autopilot across platforms I haven't even thought of.
Does that tool exist yet, or am I dreaming?
Yes, everyone on X is trying to grow their audience.
But what surprised me most is how willing people are to jump into your DMs with real advice, test your broken sandbox flows, and back your ideas.
People are literally building the tools each other need.
Who is one builder on your feed who has been genuinely helpful recently?
@devbyak It's funny what stats with us as we get older.
I've not heard of that game but it reminds me a bit of clash of clans.
I remember building my clan and joining a clan to battle others. Spent years building on that !