To expand, this would entail Claude Code (or other) generating data pipelines, data, models, training schemes, etc. ie the code agent recreates not only a SWE, but the whole of the organisation’s work over the last few years. I don’t believe access to data or compute are walls
@nandodf
-
Claude Code Generating Claude Code: Anthropic’s Future Implications
By
–
What happens to Anthropic when anyone can use Claude Code to generate Claude Code?
-
AI Safety: The Problem of Choosing Rule Systems
By
–
This video by @dwarkesh_sp is a must watch for anyone who cares about AI. Humans, corps and governments are all fallible and hence not safe AI controllers. So we could go with a set of rules, but will it be Sharia law, the US constitution, the code of Hammurabi, … ? Time and
-
Copilot Health Empowers Users to Manage Healthcare Data
By
–
Copilot Health is a great effort to empower users to manage their healthcare data and to have greater access to personalised medical expertise. Congratulations team 👏 https://t.co/SfJKe4TC5G
— Nando de Freitas (@NandoDF) 12 mars 2026Copilot Health is a great effort to empower users to manage their healthcare data and to have greater access to personalised medical expertise. Congratulations team
-
Microsoft Super Intelligence team making exciting progress
By
–
The Microsoft Super Intelligence team has been building. Exciting
-
Congratulations on Major AI Development Milestone
By
–
It’s certainly been a huge step in AI development. Congratulations
-
Open Source Data: Essential Missing Component for Large-Scale Training
By
–
More Open Source Data. The main missing ingredient for large scale training.
-
Developing New Theories with Pre-Theory Data: A Paradox
By
–
Brilliant idea. Very thought provoking. How do we develop a new theory by training a model only with data prior to the existence of that theory?
— Nando de Freitas (@NandoDF) 24 février 2026
Theory could also be a capability, like language itself or logical reasoning. https://t.co/LLsC1Q24EtBrilliant idea. Very thought provoking. How do we develop a new theory by training a model only with data prior to the existence of that theory? Theory could also be a capability, like language itself or logical reasoning.
-
Praise for Dwarkesh’s Interview with Anthropic CEO
By
–
I loved this AI interview with Anthropic’s CEO. What did I love about it? Definitely, @dwarkesh_sp’s smart and incisive questions. Great reporting! https://t.co/CQ3voISxEC
— Nando de Freitas (@NandoDF) 20 février 2026I loved this AI interview with Anthropic’s CEO. What did I love about it? Definitely, @dwarkesh_sp
’s smart and incisive questions. Great reporting! -
Coding transformation requires adaptation for millions
By
–
I fully agree. Coding has changed forever.
Millions of people will need to adapt. This is no small thing.