Tag
Anders Hejlsberg, creator of TypeScript and C#, discusses whether AI will write over 90% of code this year, noting it can handle all code for certain app classes but not high-quality code like the TypeScript compiler, and expresses hope that AI doesn't fully replace human coding.
The article highlights the hidden costs of building AI agents on external models, specifically unannounced behavioral regressions after updates that can disrupt automated workflows, and suggests strategies like version pinning to mitigate risks.
The article questions whether the focus in AI has shifted from raw LLM capabilities to agent framework engineering for real-world performance.
The article explores how agentic LLMs are redefining the role of code in software engineering, suggesting code may become a means to an end with specifications as the focus, while programming languages remain crucial for reducing ambiguity.
Kent C. Dodds defends the Model Context Protocol (MCP), arguing that it addresses a significant problem and is a worthy solution for evolution.
Un tweet que elogia a Hermes como el mejor agente autónomo actual, criticando a Elon Musk y citando a Nous Research sobre su acceso remoto y costo más bajo.
Addy Osmani and Gergely Orosz discussed engineering roles, DevTools, and AI agents in a conversation in San Francisco, with timestamps provided for a detailed discussion.
A user asks for a TL;DR on recent developments in AI companies and models, noting new models and drama, after being out of tech for a year.
Elon Musk agrees with a user's statement highlighting the benefits of robotaxis, such as saving lives, reducing costs, improving air quality, and enhancing freedom of movement.
The article discusses desired features for a modern relational query language, critiquing SQL's shortcomings and suggesting improvements based on functional programming, better syntax, and user-defined types.
Discusses the 'happy path problem' in agentic commerce and references protocols like x402, ERC 8004, ERC 7710, and A2A for handling payments, identity, permissions, and negotiation.
A discussion comparing the products of OpenAI and Anthropic, focusing on arguments for each company's AI tools like Codex, Claude, and GPT without ethical considerations.
The tweet discusses the challenge of losing proximity to builders as companies grow and emphasizes the importance of talking to users, not just customers and pipeline.
This is the ninth installment in a series examining the relationship between the NSA and the IETF, focusing on internet standards and security issues.
Gergely Orosz observes that the discussion about AI enabling mainstream formal verification has faded, and questions why AI hasn't impacted that field.
Indigo Talk speaks with the head of FDE at Cresta, discussing whether FDE is a new guise for outsourcing or a key role in AI deployment, sharing insights on AI agents, middle management, trust, and more.
Discussion on server-side technical solution selection: expose MCP or directly use SKILLS+CLI? The author believes SKILLS+CLI has become mainstream, and MCP's only advantage is the auto-updating tool discovery mechanism.
A discussion on image compression techniques or tools, shared on lobste.rs.
Tweet discusses that Python is fine for building Agents but performs poorly for Coding Agent's TUI, prompting reflection on tech stack selection.
A tweet referencing AI researcher Sebastien Bubeck suggests that certain discussed capabilities would require an advanced model like the hypothetical GPT-5.5.