Syazdev23 retweeted
this is basically the bitter lesson at work
the smartest models will need fewer and fewer tokens, they won't need to verify their work or check it for bugs, they will just be right
maybe a hot take but i now consider adversarial reviews mostly a waste of tokens. far better to use those tokens on empirical combined with formal verification. i have updated pstack to use the interrogate skill less, and eventually i think i may remove it altogether
i have found very little value with code review in the abstract, agents tend to just get lost in the nitpicks and yagnis
Syazdev23 retweeted
You can't call this AI slop anymore.
One self-taught working mom made this 9-minute AI fantasy film.
An she won 4 awards at the New York International Film Awards, including Best AI Film and Best Screenplay.
This is what real AI filmmaking looks like 👇
Syazdev23 retweeted
The Nuclear explosion is insanely good, made by Claude Opus 5.5. I ran this video in two iterations, and I used around 800k tokens, style Oppenheimer -> @lemomo_ai
"It's a fictional story of a politician losing elections and using the Infinity Gauntlet to wipe out the Earth."
Syazdev23 retweeted
Lauren Tan’s Grok Bot setup looks like a real team.
the workflow is simple:
goal → delegate → specialists work → merge results → final answer
instead of asking one agent to do everything, she splits the work across specialists:
→ Chief of Staff plans + delegates
→ Researcher finds and compares information
→ Engineer builds and tests
→ Reviewer checks quality and edge cases
→ Doc Writer turns work into reusable knowledge
→ Operator keeps workflows running
→ Memory Manager preserves context across sessions
that’s how Grok Bot starts feeling less like one assistant - and more like a small AI company working for you.
full playbook below ↓
Syazdev23 retweeted
SpaceXAI Senior Engineer Lauren Tan reveals why 99% of users are barely scratching the surface of GrokBot.
“GrokBot is the most powerful agentic tool we've ever released, yet only 1% of users know how to use it properly.”
At SpaceXAI, Lauren manages 15+ AI agents:
→ 1 Chief of Staff
→ 3 Managers
→ 11 Workers
This is how advanced AI engineering is evolving in 2026.
In just 40 minutes, she breaks down how to get more out of agentic AI and build smarter workflows with GrokBot.
More practical value than most $500 AI courses.
Skip Netflix. Watch this instead.
Save this guide for later.
Syazdev23 retweeted
this is f*cking wild. SpaceXAI engineer Lauren Tan:
"Most engineers still read every single line their agent writes
I run 60+ agents and haven't looked at their code in months. Every review note I ever left is a CI rule now"
In a 53-minute session she walks through how her agents merged 3,500 PRs last month while she was asleep
Prompting - Reviewing - Verifying - Delegating - Sleeping
Reviewing code used to be the job. Turning reviews into rules is the job now
Better than any $5,000 agent engineering course
save it, watch it, then tell me if you'd merge code you never read
Syazdev23 retweeted
Thanks to everyone who shared session IDs. We went through all of them. We found a few smaller things along the way that made some sessions burn through tokens faster than they should have, and we're rolling out fixes to those right now.
We didn't find a single root cause or any regression in Claude Code. In most cases, users who saw faster usage against limits were traced back to one or two heavy sessions (xhigh effort, /goal, Workflows, Agent Teams, long SDK jobs). These are all good ways to use Claude Code, and it's on us to make it clearer when to use them and how much these features cost.
It's good idea to keep an eye on the effort level you're on. Subagents inherit it by default and it can add up fast. /usage also shows where a session's tokens went.
If something still looks off, we'd love your /feedback. We'll also follow up directly with everyone who shared an ID.
Syazdev23 retweeted
🚨 Claude code users !!
Boris Cherny FINALLY answered the crazy usage burn !!
They found bugs making sessions burn tokens faster than normal.
Fixes are rolling out RIGHT NOW !!
Subagents INHERIT your effort level by default.
Did your limits drain faster this week ?
Syazdev23 retweeted
Don’t use Waze. Use Pintaz.
Don’t use Linktree. Use JomQR.
Don’t use Shopify. Use EasyStore.
Don’t use Eventbrite. Use Ticket2U.
Don’t use QuickBooks. Use Bukku.
Why do we need a controversy before we give Malaysian tech a chance? 🇲🇾
We keep asking when Malaysia will produce the next big tech company.
Are we willing to be their customers before they become one?
Try their products. Tell them what needs fixing. Pay for the ones that solve your problems. Recommend them.
If we only support Malaysian builders after they’ve made it, who’s supposed to help them get there?
Malaysian builders: drop your app and the tool it replaces below. Let’s grow this list.
Syazdev23 retweeted
I just cancelled 1 of my 4 Claude Max subscriptions.
Not because I don't like Opus 5.5. It's the best model I've ever used.
I cancelled because the usage is so good that I don't need 4 anymore.
OpenAI has to keep resetting limits to stop people from leaving. With Anthropic, great usage just comes with the subscription.
Syazdev23 retweeted
I'm completely floored
this is another one-shot
will need some refinement before sharing a link
I asked Opus 5.5 to explain camera focus by building an interactive lens lab
Here's what it came up with after 1 hour 26 minutes in one shot, $25.66 API cost
lens.lab.sael.net
Move the focus ring and you can see the glass elements shift the sharp plane through the scene
Syazdev23 retweeted
What? This is free?!
We rebuilt After Effects and Premiere for GPT-6 Astra and Sol.
Introducing Tesseract.
The video creative suite for your AI agent.
Give your agent direct access to our creative engine.
No more clicking through software built for people.
→ Edit footage, motion, and sound
→ Make precise changes with full creative control
→ Revise and render directly through your agent
And yes, this video was made with Tesseract.
Get the ChatGPT plugin.
Syazdev23 retweeted
What if you could learn Kuala Lumpur’s history just by playing? 🇲🇾🎮
I’m building Jalan KL with GPT 6 Astra ~ a 3D adventure where you roam the city, complete quests, and uncover the stories behind its landmarks.
For locals rediscovering home and visitors getting their first taste.
Jom jalan! 👟
try it here: jalankl.themasterofnone.xyz
Syazdev23 retweeted
GeoLibre + GeoLens: A Modern GIS Stack for Self-Hosting Geospatial Data
In this tutorial, I introduce a modern, fully open-source GIS stack for self-hosting, managing, visualizing, editing, and analyzing geospatial data using GeoLens and GeoLibre.
GeoLens provides an easy-to-deploy platform for hosting vector and raster datasets, while GeoLibre connects directly to GeoLens for interactive visualization and geospatial analysis. Together, they provide a powerful open-source workflow for organizations, researchers, and individuals who want to manage their own geospatial infrastructure.
Everything is built around open-source software and open geospatial standards.
GeoLibre: geolibre.app
GeoLibre Web: web.geolibre.app
GeoLibre GitHub: github.com/opengeos/GeoLibre
GeoLens: github.com/geolens-io/geolen…
#geospatial #opensource #MapLibre #GeoLibre #GeoLens
