Scratch

shift+tab -> /goal = ??

@xai Does this work with Scratch?

An early beta of Grok Build, an agentic CLI for coding, building apps, and automating workflows is now available for SuperGrok Heavy subscribers. Through this early beta, we will improve the model and product based on your feedback. Try it at x.ai/cli

@TheStalwart This has more to do with AI than anything else.

Even if you take the strict view that economic sentiment surveys are just political polls in disguise, it's notable that the red line here has fallen to the lowest its been during any Trump admin years.

@mattpocockuk it seems to work best when the agent is unaware of the existence of tests

Another layer of documentation I'm considering (along with CONTEXT.md and ADR's) is a list of all the agreed test seams in the app Agents simply cannot be trusted to make good decisions about what to test, and at what seam. For every small change, they extract out only what they've built into a testable function and test that. It leads to a patchwork nightmare of tests that break as soon as the implementation changes.

@comfortfajugbag we run the workflows locally now, verify the outputs in our desktop app (scratch.md), then publish straight from there 1. AI agents with skills are more flexible than n8n/make automations 2. local-file-only access makes using the AI agents safer (they can't publish)

How do you monitor if your n8n AI workflow made the right decision? With normal n8n workflows, failure is usually obvious. A node fails. An API returns an error. The workflow stops. But with AI workflows, the workflow can “succeed” while the AI still makes a bad decision. For example: wrong summary bad classification wrong lead priority weak email draft missing context wrong CRM field update That feels harder to catch than a normal error. For people building AI workflows in n8n, how are you checking quality? Do you use logs, manual review, confidence scores, Slack alerts, test data, or something else?

friendly reminder to review the output of your ai before you ship it

@OpenAI we're already to the emotional background music for a math solution video stage? the token economics must be worse than we thought

Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in 1946. For nearly 80 years, mathematicians believed the best possible solutions looked roughly like square grids. An OpenAI model has now disproved that belief, discovering an entirely new family of constructions that performs better. This marks the first time AI has autonomously solved a prominent open problem central to a field of mathematics.

@CyrusShepard i remember when the ads had a distinct background color, now it's ai (google), ads where the results used to be (google), youtube (google), more ads where the results used to be (google), maybe maps (google) unless you search for a product, then just ads (google)

Today, Google released the May Core Update. This update will classify all non-Google websites as spam. Thank you for your attention to this important matter.

brilliant concept

@charliermarsh build a real search engine

What would you do with unlimited tokens

the whole claude code vs codex debate makes it clear that neither is substantially better (higher limits != better tool)

the posthog ai agent is actually really good at creating sql reports

most of the serp is just ads anyway

@jyangballin 1. Google maps navigation 2. MyFitnessPal 3. Webflow (the platform itself) 4. Airtable 5. Supabase

Thinking about what new tasks to put in programbench v2. What software programs (CLI tool/executables? Local apps? Websites?) would u wanna see models try building from scratch?

To everyone who recently switched from Claude Code to Codex, thank you!!! Claude Code is running much faster with far higher limits now. 😁

@Ddddarren @yacineMTB what made you switch

Whew… switched over to Codex just in time. Thanks @yacineMTB

@jumperz @skyzer4ever what made you switch

@skyzer4ever could be, but almost everyone I know switched to codex.. starting with me who was on 8 months streak.. thier limit rates been a joke so..

🚨 karpathy just joined anthropic. i mean the last 6 months felt like anthropic was bleeding momentum to openai and google. you don't pull a karpathy out of retirement for small stuff. you pull him when you're about to drop something huge. could it be claude comeback? idk

@stevemordue @TTrimoreau what made you switch

@TTrimoreau Limits are not the reason I switched to Codex, and they won't be the reason I switch back when Claude becomes better

Would you still keep using codex if claude started having the same limits?

@TheAwaisManzoor @theo what made you switch

@theo switched to codex a week ago.....

Honestly I'm still really impressed with the Codex app. It works reliably. It adds useful features consistently. It has taste. The mobile integration is awesome. The git integration is solid. If you haven't used it yet, I highly recommend it.

@TheNewPolicy @theo what made you switch

@theo Yup. Anthropic runs their company like a gypsy tivoli. Switched to Codex a month ago. Never been happier.