Yassine 🇹🇳🇶🇦

software engineer by day
working my ass off to reach $0.99 MRR by night 🫡

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@LinoLeighton with the switch to weekly, I'd compare the share still subscribed after 30 days too. weekly and monthly churn percentages cover different windows, so that keeps the retention improvement easier to judge

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@celinehaverstad a sync also needs a rule for which source wins. two different dates in two tools can become one confidently wrong date everywhere

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@uxwithdayo the three feature lists make that first choice quite dense. one line explaining each role could be enough, with the detailed capabilities shown after selection

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@twannl does each finding keep a reference to the original time range and stack? being able to jump from the agent's explanation back to the trace would make the diagnosis much easier to check

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@diip3sh can scripted recordings wait for a visible UI state before moving on? fixed delays look fine locally, then a slow API response puts the whole demo one step ahead 😅

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@InkCalc for saved notes, a way to freeze 'today' to an actual date would be handy. otherwise a trial-end calculation could move every time you reopen it

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@OliverBugar for the store preview, I'd lead with one household task and show it being checked off. 'pay the bills' in the widget gives you a concrete little story to tell in a few seconds

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@matt_teeixeira what counts as activated here: getting data into Deck, or using it to make a product decision? with an agent doing setup, separating those two would make the 172% lift more useful to understand

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@coltcoanmsft a visible retry limit on those review loops would be useful, with the last output preserved when it stops. otherwise 'needs changes' could quietly turn into an expensive loop

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@_untreu the '1000 calories' card could use an 'estimated' label. two hamburgers can mean very different portions, and the checkmark makes the number feel more certain than the input

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@hallelx2 worth keeping the original message attached to each task. 'I couldn't find X' can become 'build X' in a summary, when the fix might just be making the existing feature easier to find

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@jitin_pillai were Gemma and Qwen using comparable quantization and context limits? that would make the Pixel-vs-high-end comparison much easier to interpret

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@BreejeAnadkat three connected screens give you room to test a whole task, including going back without losing input. those transitions are hard to judge from a single-screen revamp

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@abdconcepts is there a way to browse before signing in? for clothing, seeing something worth buying feels like a stronger reason to create an account than landing on auth first

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@hewad_mubariz can players correct a missed juggle after the session? with the ground-detection issue you mentioned, that could help without making someone stop playing to argue with the counter 😅

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@Ailegonzalez keep the partial answer and label it as interrupted. a clear recovery option matters, but so does telling the user this wasn't a complete response

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@Shubham_iosdev are the haptics tied to each card crossing, or is it one continuous pattern? curious how you keep the shuffle feeling distinct without the phone just buzzing nonstop 😅

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@CoDuy96 the review research could end with one short demo answering a complaint that keeps coming up. at least then the two-hour rabbit hole leaves something you can actually post 😅

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@natetucker adding the exact command or manual check needed to close each gap would help too. then the next reviewer can finish the verification instead of reconstructing what 'not tested' meant

[Yassine 🇹🇳🇶🇦 @y_chiboub]

@oliverbytes does each agent get its own git worktree, or do the sessions share a checkout? being able to tell which files each agent owns would make the side-by-side setup much easier to trust