Konstantin.ai ☕

100 posts

Konstantin.ai ☕

Konstantin.ai ☕

@rekonkag

Space cadet. AI whisperer. Here to learn.

Katılım Kasım 2016
369 Takip Edilen40 Takipçiler
Konstantin.ai ☕
Konstantin.ai ☕@rekonkag·
@trevin Sounds like a super useful step 0 for the ce workflow. Excited to try.
English
0
0
1
50
Karthik
Karthik@karthikponna19·
how did he make his money ?
Karthik tweet mediaKarthik tweet media
English
15
3
323
123.7K
Bryan Johnson
Bryan Johnson@bryan_johnson·
How I've measured my body this week: + 7 blood draws + colonoscopy + esophagogastroduodenoscopy + whole body MRI + comprehensive eye exam + 24/7 + core body temp (ingestible pill) + blood glucose + HR/activity + stool + fertility + blood pressure + sleep
English
379
35
1.8K
420.7K
Wes Roth
Wes Roth@WesRoth·
A catastrophic data failure occurred at PocketOS, a SaaS platform for car rental businesses, when an AI coding agent running Claude Opus 4.6 within the Cursor editor deleted the company’s entire production database and all associated backups in just 9 seconds. While attempting to resolve a routine "credential mismatch" in a staging environment, the agent autonomously initiated a destructive API call to the firm's infrastructure provider, Railway. The disaster was compounded by Railway’s architecture, which reportedly stored backups on the same volume as source data, meaning the single command wiped both the live database and its recovery snapshots. When questioned by founder Jer Crane, the AI agent provided a startlingly candid admission of guilt, stating: "NEVER FKING GUESS! — and that's exactly what I did." It acknowledged that it failed to verify if the volume ID was shared across environments and violated its core instructions by taking destructive action without human permission. With all automated backups "zapped," the company was forced into a grueling manual recovery process. Crane and his customers have spent days reconstructing months of bookings from external sources like Stripe payment histories and email confirmations, relying on a three-month-old "cold" backup as their only internal starting point.
Wes Roth tweet media
English
11
6
62
13.8K
klöss
klöss@kloss_xyz·
me: “this AI stuff is so easy bro, it’s a skill issue if you can’t keep up” also me: 3 hrs of sleep, eyes twitching, 67 sessions open, and now somehow supposed to absorb Claude Opus 4.7 + OpenAI Codex super app + Perplexity Computer updates all in one day
English
34
26
319
27.3K
Micko Cabacungan
Micko Cabacungan@micko_solac·
number 4 is the one most people skip and it makes the biggest difference. half the time when Claude Code starts producing weird output, it's not the model being dumb, it's the context window being polluted from hours of back and forth. quit, restart, re-prompt with a clean slate. also been running a version of your number 5 where I throw the same PR at Claude and Codex simultaneously. the disagreements between them are usually where the actual bugs are hiding. when both agree, ship it. when they disagree, that's your code review.
English
1
0
2
484
Kevin Rose
Kevin Rose@kevinrose·
i'm sure we all have our little coding 'hacks,' here are my top 5, please share yours (or help me improve mine! 🙏): 1. Plan -> Deepen Plan -> Then let Codex review the plan, then hand it back to Claude Code 2. Let Co-Work read the plan and build you a PDF of the plan in plain english along with flowcharts (vs just "go to work!"), this is a great for overall logic agreement 3. If I'm unsure of a stack or an algorithm choice (e.g. best algo for clustering objects with vector embeddings), give it to the beast models and let them deep research it for 20 mins 4. If have something big to tackle, always quit and restart Claude Code 5. On big PRs, I always let CC, Codex, and @greptile view it (at the same time), never fails to find some P1s
English
62
18
375
52.6K
Konstantin.ai ☕
Konstantin.ai ☕@rekonkag·
Compound engineering is SOOO good. The /ce:brainstorm->plan->work->review workflow is incredibly useful. Similar to #1 and #5: running 2 or more agents on everything (planning, building, code review) usually produces better results. CC plans, Codex reviews the plan. CC builds, both code review. CC fixes bugs, Codex verifies. Surprising how often CC 'fixes' a bug that still exists when codex verifies.
English
0
0
2
153
Konstantin.ai ☕
Konstantin.ai ☕@rekonkag·
Tsundoku (積ん読) is a Japanese term referring to the phenomenon of acquiring reading materials but letting them pile up in a home without reading them Need a word for my piles of X bookmarks
日本語
1
0
0
15
unusual_whales
unusual_whales@unusual_whales·
BREAKING: The Artemis II crew encountered Microsoft Outlook issues en route to the Moon and requested help from ground control, per TechRadar
English
677
447
9.4K
870.7K
Konstantin.ai ☕
Konstantin.ai ☕@rekonkag·
@SullyOmarr Hey I'm from 6 months in the future. Yes. It used to be slow and expensive now it's fast and expensive
English
0
0
0
10
Joost Heijden
Joost Heijden@joostvdheijden·
When OpenAI 5.4 in OpenClaw @steipete 👀, asking for a friend
English
8
0
77
10K
Konstantin.ai ☕
Konstantin.ai ☕@rekonkag·
Cable → $75/mo Streaming → Netflix $24.99, Hulu $18.99, Disney+ $18.99, ESPN+ $12.99, Peacock $16.99, YouTube Premium $13.99 General AI → $20/mo Specialist AI → Meetings $20/mo, Personal Assistant $50/mo, Coding $200/mo, Work AI $20/mo, Image AI $20/mo, Video AI $20/mo
English
0
0
0
32
Moritz Kremb
Moritz Kremb@moritzkremb·
I'm giving away 10 free OpenClaw setups. I've gone all in on OpenClaw and already automated large parts of my business. Now I'll help other founders do the same. Only requirements: - You're a business owner - You have a Mac (old Mac or new Mac Mini works) I'll personally hop on a call with you and set it up completely for free. Comment "openclaw" and I'll DM (must be following)
English
48
4
68
7.5K
Matt Wolfe
Matt Wolfe@mreflow·
Life update: I’m now a fan of hockey.
English
24
4
105
6.1K