Steve Tapley

Experienced first hand why raw token costs are not the whole story. Gemini 2.5 Flash vs DeepSeek V4 Flash on Open Router. Similar pricing, but Gemini chewed through 10x more tokens for the same task. And DeepSeek's results were far superior. D19 #100DaysOfCode #buildinpublic

Built & now executing the plan to address user feedback: - home page - unclear hero (fixed) - removed coming soon products (done) - unclear process (partially done) - lack of testimonials (underway) - sample output (underway) D17/18 #100DaysOfCode #buildinpublic

Toying with purchasing one of the surface scan security apps that are out there. Then I found Mozilla's HTTP Observatory site. Got both my marketing site and app site from 50/100 to 115/100 (bonus points???) It's not everything, but its free and fast.

Spent some more time on user research, and got some great, hard to hear feedback. It confirms some things, which is good because now its a plan of action. And then I spent time fixing up an auth bug, which stopped people logging in. Sigh. 🛠️ D16 #100DaysOfCode #buildinpublic

Think about the last time someone on your team let a commitment drop. Were you actually mad about the impact on the work, or the sudden radio silence? trykanso.app

Brené Brown nails the root cause: you can’t hold someone accountable for something they didn't explicitly agree to. Most "accountability issues" aren't execution failures. They're just messy, unspoken expectations.

Accountability is the one cultural value nobody admits they don't want. Yet when we looked at early data from Kanso, only 20% of people rated it as essential. 60% put it squarely in the middle: 'moderately important'.

Continuing to empty my brain into the Kanso project repo. After a fortnight of this, Codex is finally starting to 'get it'. Way more insight into what it is I need to do (lets be honest, its marketing for me right now!). D15 #100DaysOfCode #buildinpublic

Weekend was spent cleaning up articles I had previously written (ahem, generated). The quality generally sucked, so a rewrite was in order. The bones were there, the voice wasn't. Result was a cleaned up framework, style and writing guides. D14/15 #100DaysOfCode #buildinpublic

@chams_builds People who say that don't understand the builder mindset, or the craft itself.

people ask why i'm building software at 43. the startup answer doesn't fit. so here's the honest one: 1) proof to myself. 20 years inside other people's products. wanted to know if i could build my own. 2) something to show my kids; that you're never stuck to any path, until you believe so. 3) the satisfaction of making the thing. the startup "why" is a market story. the mid-career "why" is quieter. mine is: i wanted to know if i could. i still do. 🔍

Grunt work day. Package updates & patches Checking Sentry & PostHog logs. Its not glamourous work, but after a long week, doing that is a cheap dopamine hit. D13 #100DaysOfCode #Solopreneur #BuildInPublic

He kindly sent me his email address and Reddit username too. What a guy!

Interesting day today. Got a "shakedown" from a security researcher who found a "critical vulnerability" on my site. A lax DMARC rule 💀 Apparently the automated scan they ran is worth a $100 payout. The beg-bounty hustle is real. D12 #100DaysOfCode #buildinpublic #indiehackers

@chams_builds Nice, only been following for a while, but you post consistently interesting stuff

1500 of you now 🎉🙏🏻 i started posting on X as a 43yo guy with a full time job, two kids, and an itch to build something for myself; to prove myself that i can do it, and to show my kids that anything is possible. most of you found me because you're somewhere in a similar place in your own unique journey. years of experience, a stable seat, and a quiet question about whether there's still room to build something of your own. there is.

Trying to figure out distribution. I know its early days, but what do you think is wrong with the thread here? Lack of audience? Not enough time? Not well written? Too AI? Day 12 of #100DaysOfCode #buildinpublic

@jckwhls Yeah, I'm there. MVP done, marketing started... not really sure what it is I'm missing yet (because it all takes a long time to bear fruit). Working through SEO and Reddit as two main channels, X as experimental. trykanso.app

Founders, une vraie question. On passe des mois sur le produit, et la partie la plus dure arrive juste après : faire en sorte que les gens le trouvent. Moi je construis en public un outil qui dit aux fondateurs si l'IA les recommande à leurs clients, et quoi corriger si ce n'est pas le cas. Mais mon vrai défi du moment ce n'est pas le code, c'est la distribution. Et je crois qu'on est nombreux dans ce cas. Du coup j'aimerais échanger avec d'autres qui sont au même endroit : tu construis quoi en ce moment ? ton canal d'acquisition numéro 1 aujourd'hui, c'est quoi ? Je réponds à tout le monde. Et si tu veux voir ce sur quoi je bosse : ismylandingpagevalid.com On avance mieux à plusieurs.

@danmartell Do you use sample text to give it that voice? I find it helps, but I'm still never really happy with it. Maybe I need to add a second review phase to force it to clean up its mess?

Typical day writing with AI: - explain to Claude the article I want (2 minutes) - Claude writes the first draft (5 minutes) me: "why does it sound like slop?" Claude: "You're right to push back! Not because it's wrong. Not because it's bad writing. Because you're quietly getting tired of these patterns."

@BratDotAI I don't think any active project could ever be considered finished!

What’s harder with AI agents? - Starting projects - Finishing projects - Maintaining projects - Explaining projects

When you last had real autonomy, did it make your work better, or did you just stop asking for help? #workplaceculture

This is why we built Kanso. Career alignment isn't about chasing absolute freedom - its about mapping where "more of this" and "less of that" actually work for you.

In Kanso's early data, 0% of people rated autonomy as unimportant. Zero. The only factor with no dissenters.