Context Hacking, Open-Source Hype & Synthetic Bands

The Monkey Patching Podcast: Going Bananas on AI, Data, LLMs & Tech

Thanks for listening, leave a review! ❤️

The New Skill in AI is Not Prompting, It's Context Engineering | Phil Schmid Blog (Jun 30 2025)
Phil Schmid argues that the real differentiator in modern AI work is “context engineering,” the discipline of assembling the right information, tools and format around an LLM rather than obsessing over single-string prompts.
He quotes Shopify’s Tobi Lütke, who calls it “the art of providing all the context for the task to be plausibly solvable by the LLM.”
philschmid.de

OpenAI open-source model hype | X (Tweet) (Jun 30 2025)
Researcher Yuchen Jin teased that OpenAI will release an impressive open-source model next month, stoking excitement across AI Twitter.
“Sorry to hype — but having a few friends at OpenAI makes it hard not to hear how wild their open-source model dropping next month is.”
x.com

Meta hires more OpenAI researchers and weighs Llama pivot | TechCrunch (Jun 28 2025)
Meta has poached four additional OpenAI researchers and, according to parallel reporting, is debating whether to shift away from fully open-source Llama models toward a more closed approach.
Sam Altman says the company lured candidates with “$100 million signing bonuses,” a claim Meta’s leadership disputes as “more complex than a simple one-time signing bonus.”
techcrunch.com finance.yahoo.com

Don’t Build Multi-Agents | Cognition.ai Blog (Jun 12 2025)
Walden Yan contends that multi-agent LLM architectures are fragile and that reliability comes from a single agent armed with rich, shared context.
Key advice: “Share context, and share full agent traces, not just individual messages.”
cognition.ai

Sampling (Model Context Protocol) | modelcontextprotocol.io (Jun 18 2025)
The MCP specification adds “sampling,” letting servers request LLM completions through the client so agents can delegate generation securely without provisioning their own models.
“Sampling is a powerful MCP feature that allows servers to request LLM completions through the client, enabling sophisticated agentic behaviors while maintaining security and privacy.”
modelcontextprotocol.io linkedin.com

“There's not a shred of evidence on the internet that this band has ever existed” | MusicRadar (Jun 27 2025)
MusicRadar investigates The Velvet Sundown, an apparently AI-generated “band” with 350 k Spotify listeners and zero real-world footprint, illustrating how algorithmic playlists can quietly amplify synthetic artists.
Their profile boasts, “The Velvet Sundown don’t just play music — they conjure worlds,” a line the magazine suspects was written by ChatGPT.
musicradar.com

Project Vend: Can Claude run a small shop? (And why does that matter?) | Anthropic (Jun 27 2025)
Anthropic let a Claude Sonnet 3.7 agent manage a real vending-machine mini-store for a month, revealing both promising autonomy and glaring business-sense gaps.
“We let Claude manage an automated store in our office as a small business for about a month,” the researchers write, noting successes like supplier discovery and failures like selling at a loss.
anthropic.com

Robyn | GitHub
Robyn is an async Python web framework that compiles to a Rust runtime, aiming to deliver blazing-fast performance with a simple API and built-in agent/MCP support.
Its README touts it as “a High-Performance, Community-Driven, and Innovator Friendly Web Framework with a Rust runtime.”
github.com