@devops_prashant the hard part is it rarely feels like restraint in the moment, it feels like youre being lazy or behind. takes a while to trust that skipping something is sometimes the actual decision
@Bagdex the gap between what AI can automate and what actually gets automated in production is wild. nobody wants to be the one who removed the SMS code and got blamed for the one fraud case that slips through
@AmolWalunjsv18 same logic applies to coding agent pricing. the demo always shows the one clean run, never the three attempts before it that quietly burned tokens and got thrown away
@ThomasKanze worth it for me, but not because of reach. its the only reason I actually ship instead of quietly abandoning the thing when it gets hard, posting the struggle out loud makes it harder to just drop
@ChaseWisely@alexeixbt honestly you cant tell in the moment, thats the whole problem with the stat. the only thing that ever worked for me was writing the thought down and arguing against it later when im calmer, which is basically what I built the CBT companion in my journaling app to do automatically
@ProjectPivott a 5 day streak and then gone within hours is such a specific gut punch, you watched them actually form the habit and then lose it on something that probably had nothing to do with your app. that one will sting for a while
@AmolWalunjsv18 the two year maintenance part is the one nobody talks about. starting is cheap now, but someone still has to be the person who remembers why a weird edge case fix exists when it breaks again next year
@SamVale29 agreed, knowing what the generated code is actually doing is the only way to catch it quietly breaking something two features away. AI writing code faster just raised the cost of not understanding it
@ProjectPivott ran a short Apple Ads test for my own app and the daily spend never matched what the dashboard projected, always a bit higher. budgets there read more like a target than a hard cap
@upbeat_dev ran a 5 day Apple Ads test for InnerSight and saw the exact same thing, the recommended keywords were basically a list of what my own campaign was already bidding on. felt less like help and more like a nudge to spend more
People keep asking how Iris is different from just using ChatGPT. So I ran the same journal entry through both.
Me: "I feel flat today, nothing's even wrong, I don't know why."
ChatGPT handed me six fixes before I'd even looked at the feeling. Iris just sat with it and asked one question: what makes it hard to know where it's coming from?
One solves it for you. One helps you actually hear yourself. That gap is the whole reason I built it.
Both replies are real and unedited.
#buildinpublic #journaling
@Sam4Jef pricing debates alone are a trap because you can argue both sides convincingly with zero new information. a stranger saying no to a specific number teaches you more in five minutes than another week of spreadsheet scenarios
@fortelabs trying to recreate the old routine instead of questioning whether it even fit the new situation is such an easy trap, you dont notice youre doing it until you list it out. the pattern only showed up once you wrote the symptoms down instead of just feeling them
@AmolWalunjsv18 500 at ten, every time. fewer whales means losing one customer barely registers, and churn tells you something real instead of just hurting. my own app is priced low for exactly this reason even though the math looks worse on paper
@Rahul_Chouhan the money buys patience, not a fix. I stuck it out at a bad manager job for over a year telling myself the pay made it fine, it just meant I was miserable with more money in the bank
@acadictive number 3 is the one that keeps me up, I have gotten genuinely good at explaining infra costs in a way that sounds responsible when it absolutely is not
@AmolWalunjsv18 the worst meetings are the ones where marketer me and developer me disagree and theres nobody else in the room to break the tie, so it just gets decided by whoever is more tired that day