The overall experiment feels less “automated” as I can’t seem to stuff a whole bunch of feature requests into one task (it just completes one or two well when I do this). It’s forcing me to be more aware of what I’m building which is more work, but as a result, the code and
GENERATIVE AI
-
Codex Automates Testing and Debugging With Agents.MD
By
–
Okay http://
agents.md is very cool. I didn’t even have to set it up myself, just asked Codex to set up tests based on our code and add an http://
Agents.MD file. As a result, it runs all tests after tackling tasks and then uses this to debug before task is -

The Best AI Tools Beyond ChatGPT
By
–
Most people's AI knowledge ends at ChatGPT. If you want to stay on top of AI, you need to use the best tools. Project management: Notion ClickUp Monday Brainstorming: ChatGPT Grok Claude Presentations: Decktopus Gamma Beautiful AI Source Adam
-
Open Source Collaboration Accelerates AI Progress and Innovation
By
–
Without Llama (or PyTorch) there would have been no DeepSeek.
The point of open source is that you can accelerate progress by building on each other's work. -

Hidden Economic Knowledge in LLM Embeddings Revealed
By
–
Revealing economic facts: LLMs know more than they say This paper shows that hidden states (embeddings) of large language models (LLMs) contain rich economic information that can be used to estimate and impute economic and financial statistics more accurately than the LLMs’ text
-

LightLab: Diffusion Model Controls Light Sources in Images
By
–
LightLab: Controlling Light Sources in Images with Diffusion Models LightLab presents a diffusion-based method for precise, parametric control over light sources in a single image, enabling users to edit light intensity and color with photorealistic results. The approach
-

Maya: Open-Source Multilingual Vision Language Model
By
–
Behind Maya: Building a Multilingual Vision Language Model This paper introduces Maya, an open-source multilingual Vision-Language Model (VLM) designed to enhance performance on vision-language tasks across eight diverse languages. Maya addresses the underperformance of existing
-

MiMo-7B: Reasoning-Optimized Language Model Outperforms Larger Models
By
–
MiMo: Unlocking the Reasoning Potential of Language Model — From Pretraining to Posttraining MiMo-7B is a reasoning-centric LLM series optimized from pretraining through posttraining, achieving state-of-the-art performance on math and code tasks—even outperforming models 4× its
-
Famous Person Wore Secret Wire During Private AI Strategy Discussion
By
–
A famous person wore a secret hidden wire while speaking with me about AI, and I want to talk about it. Backstage at a recent AI event, I was talking with someone about private views on AI investments and strategy. We were both unfiltered. At one point, I joked, “We should’ve
-
Replicate Makes AI Models Easier for Code Editors and LLMs
By
–
We've made it easier for AI code editors and LLMs to use any model on Replicate. Just in time for Codex.
— Replicate (@replicate) 17 mai 2025
You can now:
– copy the page as markdown
– load directly in Claude or ChatGPT
– see an llms.txt page for any model
Here's the llms.txt for Google Veo-2:… pic.twitter.com/ZKKYr1D0JXWe've made it easier for AI code editors and LLMs to use any model on Replicate. Just in time for Codex. You can now: – copy the page as markdown
– load directly in Claude or ChatGPT
– see an llms.txt page for any model Here's the llms.txt for Google Veo-2: