Firehose

This presentation from the Superconductor team argues for moving agentic workflows off local machines and into shared, cloud-based environments. The speaker outlines a strategy for "multiplayer" engineering, focusing on keeping agent sessions persistent across interfaces (Slack, web, GitHub) and using automated bots to turn meeting transcripts and external signals into actionable code tasks.

A product marketer from Sonar explains why AI-generated code often introduces "verification debt," citing research that suggests AI-assisted productivity gains are frequently offset by a persistent increase in technical debt and security vulnerabilities. The talk advocates for automated, multi-layered code analysis to catch errors that human reviewers—prone to "rubber-stamping"—often miss.

This is a talk about "velocity sickness," a term the speaker uses to describe the burnout and lack of focus that happens when AI-driven coding leads to excessive PRs, fragmented work, and lost decision-making ownership. The presentation argues that engineers should shift their focus from implementation-heavy chat interfaces toward documentation-centered workflows to maintain team alignment and product quality.

This is a vendor pitch from Resolve AI arguing that as AI-driven coding increases production complexity, engineering teams need automated agents to handle "background" operational tasks like health checks, capacity reporting, and incident triage. The speaker positions their platform as a way to offload this maintenance work, focusing on the need for agents that combine execution capabilities with deep, evolving context about your specific infrastructure.

This video explains the limitations of standard chunk-based RAG for complex documents and proposes an alternative: using hierarchical document structure to let an AI agent navigate a file like a human would. It highlights Docling as the tool for reconstructing that structure from PDFs to enable "chunkless" retrieval.

This talk from GitHub Next explores two prototypes for future developer tooling: agentic workflows defined by natural language in Markdown, and ACE, a real-time multiplayer environment for collaborative coding. The core takeaway is that effective automation requires deterministic guardrails—like network and permission constraints—rather than relying on prompts that are susceptible to injection.

This talk is a practical audit of whether it’s better to refactor a legacy codebase manually or wait for AI agents to handle the migration. The speaker shares benchmarks showing that while modern models like Sonnet 4.6 and Opus 4.8 have drastically reduced the manual effort required for refactoring, fully automated "agentic" runs still suffer from high failure rates on complex tasks, making human-in-the-loop development the more reliable choice for now.

This video breaks down the Claude Certified Architect exam by framing its production scenarios as a series of "anti-patterns" to avoid in agentic engineering. It focuses on practical architectural constraints—such as branching on stop reasons, maintaining specialized sub-agents, and managing context isolation—rather than just prompting tips.

This talk frames the current "AI agent" trend as a transitional phase—akin to the early web—and argues that developers should look past simple chat interfaces toward "AI-native" software. The speaker draws parallels between historical computing milestones and modern infrastructure, highlighting Pipecat as an example of the real-time, multi-modal tooling required to build the next generation of interactive applications.

Saoud Rizwan argues that the "community" aspect of open source is effectively dead due to AI-driven noise and security risks, but contends that "open weights" models will survive because they offer a necessary economic alternative to the vendor lock-in of closed-source AI labs.

This panel discussion features representatives from NVIDIA, Prime Intellect, and Arcee AI arguing that open-source models provide superior reliability and control for enterprise use cases compared to closed APIs. The speakers focus on the strategic importance of owning model weights and training data to build internal "data flywheels" and optimize performance for specific, narrow tasks.

This panel discussion explores the practical realities of model compression, moving past the hype to explain how techniques like quantization actually function on modern hardware. The speakers discuss why layer-wise precision matters, the limitations of current benchmarks, and why they prioritize KL divergence over standard accuracy scores when evaluating quantized models. - Daniel Han (Unsloth) — Discusses the practical application of mixed-precision quantization to run massive models on consumer hardware. - Asma Beevi (NVIDIA) — Covers the technical implementation of NVFP4, a 4-bit float format designed to minimize accuracy loss. - Merve Noyan (Hugging Face) — Focuses on the democratization of AI through libraries like `bitsandbytes` and the ability to fine-tune models on limited VRAM. - Parth Sareen (Ollama) — Provides the perspective of running compressed models locally and the trade-offs between latency and model intelligence.

This is a brief commentary on the "token maxxing" trend, arguing that the current tendency to use as many tokens as possible is a byproduct of abstracted pricing. The speaker suggests that as users gain a better intuition for inference costs, they will naturally shift toward more discerning, efficient use of AI models.

This panel discussion explores the practical mechanics of model routing, moving beyond simple task-based delegation to discuss how to maintain context and cost-efficiency in agentic workflows. The speakers from Cognition and OpenRouter argue that effective routing requires keeping a frontier model for high-level planning while delegating execution to smaller, specialized models.

This video outlines a workflow for securing AI-assisted coding by treating agents like junior developers who require strict boundaries. It emphasizes using test-driven development (TDD) for small code batches, enforcing security rules via context files, and using external tools for verification rather than relying on the agent's own output.

This is a technical conversation about the role of vLLM as the underlying infrastructure for serving open-weight models. The guests discuss why companies are moving away from proprietary APIs toward self-hosted inference to gain control over performance, costs, and guardrails, and how this shift has turned open-source inference engines into critical enterprise software.

A clear, high-level primer on "model collapse," explaining how training AI on its own output leads to the degradation of rare information and the amplification of bias. It frames the phenomenon as a long-term engineering challenge rather than an immediate crisis, while outlining preventative strategies like data provenance and RAG.

Kenton Varda argues that modern cloud architecture prevents users from customizing software, proposing a new model where AI-generated features run as isolated "gadgets" rather than monolithic app updates. He demonstrates this using workerd to sandbox user-coded extensions, effectively treating apps like office documents that can be modified on the fly without breaking the core platform.

This is a founder-led pitch for Mariana Minerals, a company attempting to modernize domestic copper mining by replacing legacy manual processes with a proprietary software stack. It frames the venture as a national security play, focusing on their acquisition of a Utah-based operation to serve as a sandbox for their automation tools.

This interview with Nate Parrott focuses on his workflow within Claude Design, demonstrating how he uses the tool to generate rapid design variations and custom, task-specific interfaces. The conversation centers on using AI as a "junior designer" to handle high-volume iteration, rather than expecting perfect output on the first try.

This video explains the concept of Large Database Models (LDMs), a technique for performing semantic similarity searches directly within relational databases by treating rows as "sentences" and values as tokens. It outlines the process of binning numeric data and generating embeddings to allow SQL users to query for patterns without moving data to external AI platforms.

A wide-ranging conversation between Gergely Orosz and Simon Eskildsen that covers Simon's early career at Shopify, his transition to building Turbopuffer, and his pragmatic approach to engineering and startup infrastructure. It’s a candid look at the technical trade-offs of building a database on S3 and the realities of running a lean, remote-first company.

This is a high-level industry discussion about the launch of the Tokenomics Foundation, focusing on the need for standardized metrics to manage AI infrastructure costs. The conversation frames "tokenomics" as a bridge between FinOps and AI engineering, emphasizing that organizations must move beyond simple consumption tracking toward a framework that balances production costs, model efficiency, and measurable business value.

This video provides a conceptual framework for how developer roles are shifting from writing deterministic code to orchestrating probabilistic AI systems. It defines a spectrum of AI-assisted development—ranging from "vibe coding" to "agentic engineering"—to argue that human oversight remains critical as autonomy increases.

This is a technical overview of MCP Apps, a standard that allows AI agents to render interactive web components instead of plain text responses. The presenters explain how this protocol enables developers to ship branded, functional UI widgets that work consistently across any host supporting the specification.

This talk explains why the Model Context Protocol (MCP) tasks specification is difficult to implement, focusing on the distributed systems challenges of maintaining state across network failures and process crashes. The presenter demonstrates a durable invoice-processing workflow using FastMCP to show how servers can handle long-running tasks that require human-in-the-loop signals.

This talk provides a critique of current AI evaluation practices, arguing that "benchmaxxing" occurs because labs prioritize high scores on flawed, easily gamed benchmarks over real-world utility. Heiner details specific failure modes—such as data contamination, reward hacking, and poorly designed verifiers—and advocates for a shift toward high-quality human evaluation and domain-specific testing.

This is a high-level conceptual overview of why AI agents hallucinate and how to mitigate those risks through data grounding, tool-based reasoning, and human-in-the-loop design. It functions as a primer for developers and managers on why "confidence" in LLMs is often a byproduct of their training rather than a measure of factual accuracy.

This talk argues that AI security research is currently held back by "benchmaxxing"—relying on flawed, LLM-based grading oracles—and proposes instead the use of reproducible, sandboxed reinforcement learning environments with deterministic graders. The speaker, David Brumley, demonstrates that by grading models on their ability to trigger specific, real-world vulnerabilities (like those found in the V8 engine), researchers can move beyond hallucinated successes to measure genuine exploitation capabilities.

This presentation argues that current "long-horizon" benchmarks are often noisy or misleading because they rely on arbitrary time thresholds rather than environment design. The speakers emphasize that meaningful evaluation requires building environments where early decisions have cascading effects, and verifying success based on final state outcomes rather than intermediate judge scores.