Emre Arslan

┊Founder of of Entrepreneur & Vibe Coder

@dreamandstarss

[Emre Arslan @AarslanEmre]

"Hayat bir gün isteyene istediği verecektir ama çabuk ve fırsatçı davrananlardan arta kalanları"

[Emre Arslan @AarslanEmre]

We added a trust badge section to a client's product page last month. Tiny thing. Conversion rate on that page went up 11%. Their copy was already good. Their photos were already good. It was just the badges.

[Emre Arslan @AarslanEmre]

@AjeigbeAbi7220 hero section is one of those things where everyone has opinions but the actual test results are always surprising. what did the CTA copy end up being?

[Emre Arslan @AarslanEmre]

@shota7180 honestly curious which of these actually get used more than once. most "practical AI use cases" lists end up being things people try for 10 minutes and forget

[Emre Arslan @AarslanEmre]

@AiwithDharmik honestly the "stop wasting hours learning" framing is doing a lot of work here. sometimes the hours are the point.

[Emre Arslan @AarslanEmre]

@emollick yeah the overcalibrated "I find myself deeply moved by..." energy is hard to miss. feels like the model is performing thoughtfulness instead of just being useful

[Emre Arslan @AarslanEmre]

@AlphaSignalAI this is such a specific failure mode. it's not that they can't reason, they just don't loop the signal back. like they finished the task and forgot why they were doing it

[Emre Arslan @AarslanEmre]

@AlphaSignalAI curious how much these actually talk to each other vs. just arriving at similar problems independently. three labs, same week feels like convergence not coincidence

[Emre Arslan @AarslanEmre]

@patrickc never seen it either. always assumed I would at some point and just... haven't. is the book worth reading if you're not deep into art history?

[Emre Arslan @AarslanEmre]

been running the same dream interpretation prompt on gpt-4o and claude 3.5 for two weeks side by side. claude is more poetic. gpt-4o is more literal. users actually prefer the literal one. genuinely did not expect that.

[Emre Arslan @AarslanEmre]

@faisalziaanwer grew him 50% and got fired for a COD ratio shift lol. we've had clients leave over smaller things. doesn't make it less annoying though

[Emre Arslan @AarslanEmre]

@emilylai so what's the new edge then? genuinely asking. because I see people say "brand" and "community" but nobody has a clean answer for what that actually means in practice

[Emre Arslan @AarslanEmre]

@_MaxBlade $45k across three apps is solid. what's the split looking like? one carrying most of it or fairly even?

[Emre Arslan @AarslanEmre]

@Jacobsklug @Lovable 250 production apps is real signal. curious what % needed major rearchitecting after the spec phase vs ones that shipped clean

[Emre Arslan @AarslanEmre]

@aiedge_ shipped a .env once in a public repo. different era, same energy.

[Emre Arslan @AarslanEmre]

@jazzyslide what did you end up customizing first? curious whether it was the order history layout or something around subscriptions

[Emre Arslan @AarslanEmre]

@SalesforceDevs @salesforce what's the actual quality like for coding tasks? genuinely curious. every local model I've tried still fumbles on anything non-trivial

[Emre Arslan @AarslanEmre]

@jasonlk @samdblond @MonacoGTM curious how you define message-market fit in practice though. because "good message" and "message that converts cold outreach" feel like two different things to me

[Emre Arslan @AarslanEmre]

assumed our avelize clients wanted more automation.\n\nspent 3 weeks building smart reorder triggers, inventory sync, the whole thing.\n\neveryone asked to turn it off.\n\nthey didn't want automation. they wanted better visibility so they could decide themselves.

[Emre Arslan @AarslanEmre]

clients ask for a new homepage design.\nwhat moved revenue: fixing the abandoned cart email sequence.\n\nthey ask for a logo refresh.\nwhat worked: adding a size guide to the PDP.\n\nI just build what they ask. but I tell them what I found.

[Emre Arslan @AarslanEmre]

Genuine question for anyone running an AI product: do you tell users when you update the model behind it?

I haven't been. Not sure if that's the right call.