PR

[PR @rom__im]

@felixprehn GOLD

[PR @rom__im]

@consumerxai …cheap OCR on the screenshot and anchor the tap to a text landmark (headline center + fixed offset), then OCR again after the tap to verify state instead of locating buttons. zero inference in the hot path, been running it a year+ no issues

[PR @rom__im]

@consumerxai been down this road, skip the ml-per-screenshot thing. tiktok’s layout is fixed per screen size so 95% of taps can just be hardcoded normalized coords, no model needed. for the few elements that move…

[PR @rom__im]

@alexolim_ Hooks

[PR @rom__im]

#OpenClaw = PROCRASTINATION engine disguised as a dev environment.

Hours debugging crashes & hunting replacements for #Opus. I've been there too. That isn't "working".

Move the few skills and workflows that are productive to #ClaudeCode. Drop the rest.

Start shipping now.

[PR @rom__im]

Trying @WillowVoiceAI with its new Atlas 1 frontier model.

Impressed with the results. It really outperforms the other models and saves quite a lot of time doing corrections.

#Willow > #FreeFlow (Using Groq API) > #WisperFlow

Well done guys.

[Willow @WillowVoiceAI]

Introducing Atlas 1. Willow's new frontier speech-to-text model.

It outperforms ElevenLabs, Deepgram, OpenAI, and more by a wide margin.

Built on the first scalable, human-powered transcription infrastructure ever built for real-time dictation.

[PR @rom__im]

@adriansolarzz KLING

[PR @rom__im]

@AssilRachidi @davidasinclair They died from kidney failure. Probably caused by the advanced myeloma they were suffering from, the medication/chemotherapy they had already taken, or because in that trial the dose of SRT501 was 5g/day (10 times higher than the normal dose of trans resveratrol).