Yulo

Building.

[Yulo @yulo_tech]

After 9 months of discovering what "SaaS" means and having launched Vidalgo 7 months ago,

here are the 13 mistakes I will avoid in the future 👇

1/ Using NoCode: Probably one of my biggest regrets today is thinking that because I haven't been coding since birth, it would be impossible to learn while launching an app (even though it is probably the best thing to do). We ended up using Bubble and spent just as much time understanding the app as we would have learning a typical stack.

2/ We lost an enormous amount of time learning a pseudo programming language (Bubble), and because we didn’t "understand" what we were doing, we were literally unable to launch certain features that users requested. The ecosystem is also much more limited than coding, and backend functionalities offer very few possibilities, making you heavily dependent on a vast number of APIs (which destroys your margins and makes you extremely vulnerable to a complete app interruption).

3/ Another huge mistake that cost us dearly (literally) was spending an obscene amount of cash on fixed costs (API and tools). You regret it bitterly when it's time to give money to Mark for Meta Ads or just to pay some influencers.

4/ As I mentioned earlier, we had to use many APIs, especially for video generation since we couldn't use Remotion with Bubble (or at least had no idea how to integrate it). We had to significantly reduce our margin and were limited in video generation since we couldn't have pre-rendering like with Remotion and didn’t have pay-per-use but subscription-based services.

5/ Our execution was disastrous. This is probably what cost us the most out of everything. We spent an infinite amount of time deciding whether to do this or that. No prioritization of important tasks, constantly seeking perfection. And then suddenly, we wake up and realize that boom, almost 8 months have passed, and we're far from achieving what we set out to do. Just to tell you, we had the idea to launch Vidalgo in December 2023 and released a beta version around February, well before a huge number of other competitors arrived (who did very well, by the way). They all came well after us but surpassed us by moving extremely quickly while we were literally struggling. Well played to them!

6/ Not defining our ICP. A very classic mistake, but I guess everyone goes through it. It cost us an infinite amount of time concentrating our energy on things that simply did not target our actual user base. Never again will we make this mistake. It's simple yet saves a huge amount of time. You stop talking to everyone foolishly, thinking you can reach 8 billion humans.

7/ Wanting to "innovate" in terms of UI. That's over for us as well. We learned the hard way that we are not designers and that from now on, it will be very simple: UI kit + designs that have already proven themselves. That's it. Too much mental energy wasted unnecessarily on a blank Figma page.

8/ Speaking of Figma, one thing I won’t do again is jumping into creating the interface for real, thinking I have a "brilliant plan in my head," only to spend hours and hours modifying everything, hoping to stumble upon the perfect design. Always start with a Figma mockup in the future before beginning.

9/ NOT TALKING TO YOUR CUSTOMERS!!! You might think this is another silly mistake and that everyone keeps repeating it. And yes, you're right! We did it very early on but quickly lost the habit, and it showed in the product. We started making modifications and releasing features that were not at all relevant. Even a small message from time to time to some customers on Discord, for example, doesn't hurt. And given our execution problem, I might place this mistake as equally important. The worst part is that it seems so cliché, but I think by doing completely useless things, we forget to focus on the fundamental pillars.

10/ Not directly related to the product itself, but not having installed @rize_io and an extension to block X and YouTube until 5 pm also cost us a lot of time. (Rize is really amazing; I'm literally in love with the product.)

11/ Paying for AI tool directories. While it might be a good idea, for us, honestly, aside from wasting money, it didn't do anything significant. I think it mainly depends on the product in the end, but I don’t think it’s the most worthwhile investment at the beginning (especially for expensive paid directories).

12/ Underestimating the power of organic content (TikTok, Shorts, Reels...) was also a fatal mistake for a B2C product like ours. In fact, for a very long period, we did the exact opposite of what we should have done in marketing by focusing more on B2B channels instead of B2C. It’s very stupid, I know, but oh well 🤷‍♂️

13/ Creating a full AI product, especially for our very first SaaS, is clearly suicidal ahaha 😅. The number of APIs and agents we have to manage, models becoming temperamental overnight and needing to be changed quickly, the new image models with their very strange prompting rules that need to be understood, in short, it’s a nightmare to manage. Moreover, all these little APIs have a cost that significantly reduces your margin, especially at the beginning when you stay on some very expensive entry-level plans (Elevenlabs...). And the "random" results are also difficult to tame while doing everything possible to have something relatively stable and functional.

----> All these cumulative mistakes made us miss an excellent opportunity and momentum that was right in front of our eyes.

But in the end, the most important and precious thing for us is all that we have learned during this journey. This product has mainly provided us with valuable resources that are now deeply ingrained in us and that we will be able to use for every new product. It has brought us much more than any training or other similar experiences. We learned everything from scratch about marketing and development thanks to this wonderful community, which is of inestimable value and constantly provides value daily. Thank you, indie hackers! 🤝

[Yulo @yulo_tech]

@totoche Yes j'utilise ça actuellement

[Yulo @yulo_tech]

@FlorinPop17 👋

[Yulo @yulo_tech]

@s_chiriac Will you replace her with an AI agent?

[Yulo @yulo_tech]

@posthog When can we try it?!

[Yulo @yulo_tech]

Don't make landing pages with AI

I will never buy a product that does this because it just means the team doesn't care about everything else

[Yulo @yulo_tech]

@jackfriks Why not buy a Whoop?

[Yulo @yulo_tech]

@tibo_maker Are you using claude to build it ?

[Yulo @yulo_tech]

@vascoabm You can use @chargeblast if you want to avoid them

[Yulo @yulo_tech]

@SaidAitmbarek @melvynx Maybe Codex isn't that cheap after all

[Yulo @yulo_tech]

@SaidAitmbarek it's true?

[Yulo @yulo_tech]

@adrianowhoscale ça track google ads aussi ?

[Yulo @yulo_tech]

@0xishang When will there be a Chargeblast integration?

[Yulo @yulo_tech]

@melvynx I prefer using Opus 4.6 as it consumes fewer tokens than 4.7 and 4.8 due to the tokenizer change

So it might be worth comparing with previous versions as well

[Yulo @yulo_tech]

@Trendtrack_io 👏

[Yulo @yulo_tech]

@melvynx "Refactor everything in ruby, please make no mistakes"

[Soraia @SoraiaDev]

MiroMiro is live on Uneed today!

would love your upvote if you like it 🫶

[Yulo @yulo_tech]

@Proton_Pass You forgot 30+ breaches in France 😂

[Yulo @yulo_tech]

@asklumo Love Proton but Lumo sucks, It just doesn't work

I feel like I'm using GPT-3

[Yulo @yulo_tech]

@brivael where are you going?