Turn Feedback into Customer Loyalty with SeggWat
Turn Feedback into Customer Loyalty with SeggWat
Today I tried a little experiment: Claude remote control on my cell phone β in the middle of the Black Forest.
While hiking, I implemented a major feature for SeggWat. Sounds crazy, but it worked amazingly well.
Now I'm sitting at my desk doing the code review. And yes, the code looks solid.
Our Chrome extension sits in your side panel. See a bug? Click, screenshot, annotate, submit.
Done.
Your team stays in flow. Your product team gets structured feedback with full context.
I finally got an overview of my infrastructure.
I built a Mintlify-like docs framework for Dioxus. MDX content embedded at compile time β no runtime file I/O, no separate build step. Just dx serve.
dioxus-docs-kit.oxidt.com
What do you built with dioxus?
Thx for your feedback :) For feature voting @fider.io is great. Also adding this ability to @seggwat.bsky.social right now
Still not too late! What are you using? Would love to hear what features matter most to you. βοΈ
First product: 3 months.
Second product: 1 month.
Shipping is a skill. Every product you build makes the next one better. You know which corners to cut and which to protect.
You build the MVP, not the fantasy.
The best shortcut? Start.
#buildinpublic
We are the thin generation between "I can't find the internet" and "what's a file system?"
We teach upward. We teach downward. Digital translators for both ends of the timeline. There will never be another generation like us.
Sentry alert hits. You open the browser. Read the stack trace. Switch to your IDE. Find the file. Write the fix.
Or: "Fix the TypeError in production."
Sentry's new CLI lets your AI coding agent query errors, read stack traces, and propose fixes β without leaving the terminal.
Spent more time evaluating hosted auth than I'd like to admit.
Every provider: great docs β quick setup β custom login screens? -> That's the enterprise plan.
Now: custom flow with Zitadel. Self-hosted, open source, every pixel mine.
Best detour I ever took.
#buildinpublic
"We didn't change anything"
checks deployment
image tag: latest
You did. You just don't know what.
AI-assisted coding is the digital photography moment for software development.
Companies that skip it won't just be slowerβthey'll be outpaced by teams shipping 10x faster.
The gap is widening. Every. Single. Day.
Genuine question for devs using AI assistants:
How often does Claude decide your auth system is "unnecessary complexity" and just... removes it?
Asking for a friend.
Hot take: most product teams collect feedback in Slack threads, spreadsheets, or not at all β then wonder why they ship the wrong features.
We're building a tool to fix that. No guessing what you need β you tell us.
What's the #1 feature that would make you actually switch? π
Yeah, agent-browser works faster for me π
Writing e2e tests with agent-browser in Claude Code is surprisingly smooth.
Describe what you want to test in plain language β AI navigates, clicks, fills forms, and validates.
No more brittle selectors. Tests that read like requirements.
Love devcontainers much, but use also often docker compose .
Building my own affiliate platform vs paying for an existing solution... The dev in me wants to build, but is it worth it?
What's your take?
#BuildInPublic
Curious builders, what's your local dev setup?
π³ Docker Compose
π¦ Devcontainers
βοΈ PaaS from day one
π οΈ Bare metal chaos
Drop your stack belowβbonus points for why you switched π
#DevOps #BuildInPublic
The calculus changed.
AI handles the complexity. You get the benefits.
Maybe it's time to reconsider which languages we ask AI to write.
#RustLang
π° Better economics
β’ Superior performance = fewer compute hours
β’ No GC pauses = predictable latency
β’ Consistent resource usage = easier capacity planning
When scaling, efficiency compounds.
π Memory safety without garbage collection
Rust eliminates entire categories of bugsβbuffer overflows, use-after-free, data races.
No runtime overhead. No safety trade-offs.
AI just made Rust's biggest criticism irrelevant.
The knock on Rust? Steep learning curve. Ownership. Borrowing. Lifetimes.
But if AI writes the code, that doesn't matter anymore.
What's left are pure advantages π§΅
Am I the only one who creates new accounts because I forgot which email I used last time? π
Currently using email auth for my SaaS. Thinking of switching to passkeys.
What do you prefer?
45 days live. 0 customers.
The honest #buildinpublic update for
@seggwat.bsky.social:
β
Product works
β
Docs written
β
Widgets ready
β Distribution figured out
Building is the easy part. Finding customers is the real game.
Back to it.
#BuildingInPublic
How it works:
- Claude Code's native skills system loads at startup
- This skill writes new skills when it finds extractable knowledge
- Descriptions optimized for future retrieval
Check it out: github.com/blader/claud...
When Claude discovers something non-obvious during debugging, it saves that knowledge as a new skill.
Next time a similar problem comes up, the skill loads automatically.
Persistent learning across sessions.
Every AI coding session starts from scratch.
You spend an hour debugging some obscure error. Claude figures it out. Session ends.
Next time you hit the same issue? Start over.
This Claude Code skill changes that π§΅