Hi! There isn't a lot of activity recently from my side. I'm would close that demo because no-one is using it. Yes, it's a harsh truth that only through spending few hundreds euros on ads I managed to make random users from internet to try the app. Thread
I'm a developer and as everyone on twitter is amazed by how easy it to create a pet project now. Mistake #1: I haven't done any user validation. Tech stack: firstly in Lovable then in Cursor for further polishing through Github and one-click deployment hosting from Netlify
I created this account to try to receive user's attention to my product and to understand how hard is it in reality road from 0 to 1 dollar/euro. Mistake #2: don't start replying/posting if you are not ready to do that for months without any user feedback at all - I was not.
Idea was the following - another ai generator but with unique styles. Users would like those and pay for batch of 50 images to generate more. How does it work from tech AI perspective? I used to fine-tune that particular model ...
by just copying cool images that I found on twitter which were generated from tools like Midjourney. There are quite a few accounts here who are doing that - posting a cool pic with a --seed that you can replicate in Midjourney. I took such images and trained LoRAs models on them
made that process simple - I uploaded a few images, paid some bucks to start training - and in up to 5 minutes you can generate images in that particular style. And a as developer you can use API from here and receive images
Fairly simple, isn't it? So I took images that I like and added them as styles so that user can choose and without any --seed or subscription to Midjourney obtain images in selected styles. What about pricing, how to charge users? It's not that straightforward
For each API call I'm paying a few cents plus one-time few bucks for each new style. Plus costs of ads! So in your price you should reflect those points in order to make it break even to be at least tiny profitable. In my case, on 50 images I charged 5 euros. How much you think
is pure profit here(made 0)? In my case it should be around 10% percent: cost per API call for 50 images + ads + taxes let's omit training Loras. Ads is a huge part in expenses. I spent hundred of euros in Google Ads to receive ~100 user generated images to look at. How?
I have a @supabase DB for it, which allows me to see which style user selected, prompt and generated image. Surprise-surpise! Mistake #3. At least in image generation domain be prepared that users would try to generate NSFW content. And be prep to deal with it
After having proper filtering for NSFW adding more styles I rested - app is functional. Paid some money to tiny ads from for one month, for launch (one-time), and google ads. in google ads you can
for feedback I created simple form in Google because I tried one platform for feedback and it was too overcomplicated back then. It's a button with feedback here . And it was "cool app!", "ahsdbasg", "bad". Mistake #5: User feedback would be mostly useless
see audience and platform - so that were I saw that product was used mostly from mobile (85%) and I rewrote design with Lovable/Cursor to look better here. Mistake #4: track how user is using your product from the start. tools like @Superwall might be cool. I had @Stripe just
So in the end it's all pure negative experiment from financial points of view - lost few hundreds euros intentionally and received some experience back. I would shut down app somewhere this month. Oh, and yes, also important one:
When you starting a new project distinguish between "it's cool a want to build and earn money from it" or "it's cool from tech perspective, I would learn a new concepts that would make me a better engineer, zero intention to make any money from it". There are not the same!
Mistake #6: I build generic tool image generation with unique styles which noone cares like many other apps. Create unique proposition, go into niche topic, check is it needed/solves a problem that someone would pay for!! first before building it.
If you building to learn, then skip those above and just choose technologies and build with them to gain more knowledge with those. Even use Kubernetes for that, who cares. If build to sell - be ready that it might not be fascinating and not interesting from tech perspective
But anyway, thanks for all the posts from diff build-in-public creators like @levelsio and similar who inspired me to actually try and fail successfully I would say. In the end, that knowledge is more valuable for me then all posts like this "I make 120K MRR, started yesterday"
Why google ads? I tried other platforms. Reddit ads @Reddit want to have a call with me and I thought I'm too small and that experiment is too small(I haven't responded to them, sry). But nice experience, they actually tried application by registering, really thanks for that!
x ads are disgusting! I didn't manage to give them money and click on button "give me users" - it's much harder in reality with all verifications steps, campaign creation, etc. As a developer, setting up ads is experience in itself and quite a diff topic on it's own I would say.
And check vibe-coding security guys! In case of using @supabase and @lovable I found out that actually my DB design was so bad, that free user can just adjust credits assign to him/her and keep using product permanently without paying any 50 images pack at all. Do RLS!
TLDR: sent some hundreds in money and almost a year with occasional coding/marketing and received some experience from it. Huge respect to all sales and marketing, SEO people - it's a tough job to get user/google search attention to your product.
P.S.: for SEO - vibe-coding app are usually SPA which means they renders as single HTML page. Lovable is using Vite for example. So Google rank them bad. Consider rewriting your app to be SSR - with server side rendering, like react router, add real backend with next.js etc.
P.S.: I haven't rewrote app to it, so still SPA for now so can't provide own experience on that topic. Hope that's would be useful for you, occasional twitter reader
Hi! There isn't a lot of activity recently from my side. I'm would close that demo because no-one is using it. Yes, it's a harsh truth that only through spending few hundreds euros on ads I managed to make random users from internet to try the app. ThreadI'm a developer and as everyone on twitter is amazed by how easy it to create a pet project now. Mistake #1: I haven't done any user validation. Tech stack: firstly in Lovable then in Cursor for further polishing through Github and one-click deployment hosting from NetlifyI created this account to try to receive user's attention to my product and to understand how hard is it in reality road from 0 to 1 dollar/euro. Mistake #2: don't start replying/posting if you are not ready to do that for months without any user feedback at all - I was not.Idea was the following - another ai generator but with unique styles. Users would like those and pay for batch of 50 images to generate more. How does it work from tech AI perspective? I used to fine-tune that particular model ...by just copying cool images that I found on twitter which were generated from tools like Midjourney. There are quite a few accounts here who are doing that - posting a cool pic with a --seed that you can replicate in Midjourney. I took such images and trained LoRAs models on themmade that process simple - I uploaded a few images, paid some bucks to start training - and in up to 5 minutes you can generate images in that particular style. And a as developer you can use API from here and receive imagesFairly simple, isn't it? So I took images that I like and added them as styles so that user can choose and without any --seed or subscription to Midjourney obtain images in selected styles. What about pricing, how to charge users? It's not that straightforwardFor each API call I'm paying a few cents plus one-time few bucks for each new style. Plus costs of ads! So in your price you should reflect those points in order to make it break even to be at least tiny profitable. In my case, on 50 images I charged 5 euros. How much you thinkis pure profit here(made 0)? In my case it should be around 10% percent: cost per API call for 50 images + ads + taxes let's omit training Loras. Ads is a huge part in expenses. I spent hundred of euros in Google Ads to receive ~100 user generated images to look at. How?I have a @supabase DB for it, which allows me to see which style user selected, prompt and generated image. Surprise-surpise! Mistake #3. At least in image generation domain be prepared that users would try to generate NSFW content. And be prep to deal with itAfter having proper filtering for NSFW adding more styles I rested - app is functional. Paid some money to tiny ads from for one month, for launch (one-time), and google ads. in google ads you canfor feedback I created simple form in Google because I tried one platform for feedback and it was too overcomplicated back then. It's a button with feedback here . And it was "cool app!", "ahsdbasg", "bad". Mistake #5: User feedback would be mostly uselesssee audience and platform - so that were I saw that product was used mostly from mobile (85%) and I rewrote design with Lovable/Cursor to look better here. Mistake #4: track how user is using your product from the start. tools like @Superwall might be cool. I had @Stripe justSo in the end it's all pure negative experiment from financial points of view - lost few hundreds euros intentionally and received some experience back. I would shut down app somewhere this month. Oh, and yes, also important one:When you starting a new project distinguish between "it's cool a want to build and earn money from it" or "it's cool from tech perspective, I would learn a new concepts that would make me a better engineer, zero intention to make any money from it". There are not the same!Mistake #6: I build generic tool image generation with unique styles which noone cares like many other apps. Create unique proposition, go into niche topic, check is it needed/solves a problem that someone would pay for!! first before building it.If you building to learn, then skip those above and just choose technologies and build with them to gain more knowledge with those. Even use Kubernetes for that, who cares. If build to sell - be ready that it might not be fascinating and not interesting from tech perspectiveBut anyway, thanks for all the posts from diff build-in-public creators like @levelsio and similar who inspired me to actually try and fail successfully I would say. In the end, that knowledge is more valuable for me then all posts like this "I make 120K MRR, started yesterday"Why google ads? I tried other platforms. Reddit ads @Reddit want to have a call with me and I thought I'm too small and that experiment is too small(I haven't responded to them, sry). But nice experience, they actually tried application by registering, really thanks for that!x ads are disgusting! I didn't manage to give them money and click on button "give me users" - it's much harder in reality with all verifications steps, campaign creation, etc. As a developer, setting up ads is experience in itself and quite a diff topic on it's own I would say.And check vibe-coding security guys! In case of using @supabase and @lovable I found out that actually my DB design was so bad, that free user can just adjust credits assign to him/her and keep using product permanently without paying any 50 images pack at all. Do RLS!TLDR: sent some hundreds in money and almost a year with occasional coding/marketing and received some experience from it. Huge respect to all sales and marketing, SEO people - it's a tough job to get user/google search attention to your product.P.S.: for SEO - vibe-coding app are usually SPA which means they renders as single HTML page. Lovable is using Vite for example. So Google rank them bad. Consider rewriting your app to be SSR - with server side rendering, like react router, add real backend with next.js etc.P.S.: I haven't rewrote app to it, so still SPA for now so can't provide own experience on that topic. Hope that's would be useful for you, occasional twitter reader
Hi! There isn't a lot of activity recently from my side. I'm would close that demo because no-one is using it. Yes, it's a harsh truth that only through spending few hundreds euros on ads I managed to make random users from internet to try the app. Thread ... I'm a developer and as everyone on twitter is amazed by how easy it to create a pet project now. Mistake #1: I haven't done any user validation. Tech stack: firstly in Lovable then in Cursor for further polishing through Github and one-click deployment hosting from Netlify ... I created this account to try to receive user's attention to my product and to understand how hard is it in reality road from 0 to 1 dollar/euro. Mistake #2: don't start replying/posting if you are not ready to do that for months without any user feedback at all - I was not. ... Idea was the following - another ai generator but with unique styles. Users would like those and pay for batch of 50 images to generate more. How does it work from tech AI perspective? I used to fine-tune that particular model ... ... by just copying cool images that I found on twitter which were generated from tools like Midjourney. There are quite a few accounts here who are doing that - posting a cool pic with a --seed that you can replicate in Midjourney. I took such images and trained LoRAs models on them ... made that process simple - I uploaded a few images, paid some bucks to start training - and in up to 5 minutes you can generate images in that particular style. And a as developer you can use API from here and receive images ... Fairly simple, isn't it? So I took images that I like and added them as styles so that user can choose and without any --seed or subscription to Midjourney obtain images in selected styles. What about pricing, how to charge users? It's not that straightforward ... For each API call I'm paying a few cents plus one-time few bucks for each new style. Plus costs of ads! So in your price you should reflect those points in order to make it break even to be at least tiny profitable. In my case, on 50 images I charged 5 euros. How much you think ... is pure profit here(made 0)? In my case it should be around 10% percent: cost per API call for 50 images + ads + taxes let's omit training Loras. Ads is a huge part in expenses. I spent hundred of euros in Google Ads to receive ~100 user generated images to look at. How? ... I have a @supabase DB for it, which allows me to see which style user selected, prompt and generated image. Surprise-surpise! Mistake #3. At least in image generation domain be prepared that users would try to generate NSFW content. And be prep to deal with it ... After having proper filtering for NSFW adding more styles I rested - app is functional. Paid some money to tiny ads from for one month, for launch (one-time), and google ads. in google ads you can ... for feedback I created simple form in Google because I tried one platform for feedback and it was too overcomplicated back then. It's a button with feedback here . And it was "cool app!", "ahsdbasg", "bad". Mistake #5: User feedback would be mostly useless ... see audience and platform - so that were I saw that product was used mostly from mobile (85%) and I rewrote design with Lovable/Cursor to look better here. Mistake #4: track how user is using your product from the start. tools like @Superwall might be cool. I had @Stripe just ... So in the end it's all pure negative experiment from financial points of view - lost few hundreds euros intentionally and received some experience back. I would shut down app somewhere this month. Oh, and yes, also important one: ... When you starting a new project distinguish between "it's cool a want to build and earn money from it" or "it's cool from tech perspective, I would learn a new concepts that would make me a better engineer, zero intention to make any money from it". There are not the same! ... Mistake #6: I build generic tool image generation with unique styles which noone cares like many other apps. Create unique proposition, go into niche topic, check is it needed/solves a problem that someone would pay for!! first before building it. ... If you building to learn, then skip those above and just choose technologies and build with them to gain more knowledge with those. Even use Kubernetes for that, who cares. If build to sell - be ready that it might not be fascinating and not interesting from tech perspective ... But anyway, thanks for all the posts from diff build-in-public creators like @levelsio and similar who inspired me to actually try and fail successfully I would say. In the end, that knowledge is more valuable for me then all posts like this "I make 120K MRR, started yesterday" ... Why google ads? I tried other platforms. Reddit ads @Reddit want to have a call with me and I thought I'm too small and that experiment is too small(I haven't responded to them, sry). But nice experience, they actually tried application by registering, really thanks for that! ... x ads are disgusting! I didn't manage to give them money and click on button "give me users" - it's much harder in reality with all verifications steps, campaign creation, etc. As a developer, setting up ads is experience in itself and quite a diff topic on it's own I would say. ... And check vibe-coding security guys! In case of using @supabase and @lovable I found out that actually my DB design was so bad, that free user can just adjust credits assign to him/her and keep using product permanently without paying any 50 images pack at all. Do RLS! ... ... TLDR: sent some hundreds in money and almost a year with occasional coding/marketing and received some experience from it. Huge respect to all sales and marketing, SEO people - it's a tough job to get user/google search attention to your product. ... P.S.: for SEO - vibe-coding app are usually SPA which means they renders as single HTML page. Lovable is using Vite for example. So Google rank them bad. Consider rewriting your app to be SSR - with server side rendering, like react router, add real backend with next.js etc. ... P.S.: I haven't rewrote app to it, so still SPA for now so can't provide own experience on that topic. Hope that's would be useful for you, occasional twitter reader
Missing some Tweet in this thread? You can try to
Update
Unroll Another Thread
Convert any Twitter threads to an easy-to-read article instantly
Have you tried our Twitter bot?
You can now unroll any thread without leaving Twitter/X. Here's how to use our Twitter bot to do it.
