The Hard Truth About Building AI Agents
The speaker warns that building agentic AI requires careful, selective instruction design because developers cannot embed every rule or constraint into prompts. Even with very large context windows, empirical limits mean only a fraction of tokens should be used effectively, so overloading prompts diminishes performance. Engineers must balance token cost, model capability and context utility to achieve reliable behavior rather than attempting to encode exhaustive directives. This creates a practical ceiling on how much instruction-driven control can be exerted over agents.
“I’ll Burn Out 2 in Minutes” The Brutal Reality of GPU Clusters
The speaker describes stress-testing new GPU clusters by immediately pushing them to maximum load and routinely causing about 2% of units to fail within minutes because many accelerators are engineered to run extremely hot and rely on substantial cooling. He...
Write Reliable Software with Temporal
The video introduces Temporal’s durable execution model as a way to boost developer productivity when building agentic systems. It explains how Temporal abstracts reliability concerns, allowing developers to write ordinary code that runs to completion despite cloud‑scale failures, flaky services,...
MLOps Coding Skills: Bridging the Gap Between Specs and Agents
The article introduces Agent Skills, a lightweight markdown‑based tool that injects organization‑specific engineering standards into AI coding agents. By converting sections of the MLOps Coding Course into SKILL.md files, the author shows how agents can automatically apply preferred tools such...
Using Agents in Production: Past Present and Future // Euro Beinat
Prosus announced it has shipped nearly 8,000 AI agents, with only 15% achieving production status while the remainder function as learning experiments. The data was presented at the Computer History Museum’s Coding Agents virtual conference on March 3, where industry...
Context Engineering 2.0: MCP, Agentic RAG & Memory // Simba Khadder
Simba Khadder unveiled Redis Context Engine at the Coding Agents Conference, positioning it as "Context Engineering 2.0" that merges retrieval, tool invocation, and memory into a single MCP‑native surface. The platform treats documents, databases, events, and live APIs as addressable resources via...
Enterprise-Ready MCP // Jiquan Ngiam
The Coding Agents Conference on March 3 will feature Jiquan Ngiam discussing the rapid enterprise adoption of agents and Model Context Protocols (MCPs). Over 80 % of professional developers now use AI tools daily, and agentic coding platforms such as Claude Code...
MLflow Leading Open Source
Databricks’ leaders Corey Zumar, Jules Damji, and Danny Chiao discussed the latest evolution of MLflow on the MLOps Podcast. The open‑source platform is being rebuilt to handle generative AI, agent workloads, and production‑grade governance, moving beyond its original data‑science‑only focus....
Simulate to Scale: How Realistic Simulations Power Reliable Agents in Production // Sachi Shah
At the Computer History Museum’s Coding Agents Conference, Sachi Shah presented how realistic, scalable simulations are essential for deploying reliable AI agents in production. She explained that simulations can mirror messy real‑world interactions—including multilingual dialogue, emotional states, background noise, and...
Yes, We Do Need MCP
The upcoming Coding Agents Conference will feature a deep‑dive into MCP, a stateful communication protocol designed for AI agents. Organizers argue that MCP’s built‑in statefulness differentiates it from gRPC and HTTP, enabling conversations to resume after interruptions. The talk will...
Building an Orchestration Layer for Agentic Commerce at Loblaws
The talk introduced Alfred, Loblaws’ production‑grade orchestration layer designed to power agentic commerce across its massive retail ecosystem. Built on Google Kubernetes Engine with a FastAPI gateway, Alfred abstracts LLM providers, leverages LangChain‑style execution graphs, and connects to over fifty...
Agents as Search Engineers // Santoshkalyan Rayadhurgam
Santoshkalyan Rayadhurgam argues that the foundational assumption of classic retrieval—users supply fully formed intent—is collapsing, prompting a transition from deterministic, stateless pipelines to agentic, stateful search systems that reason across turns. He contrasts three generations: lexical BM25 pipelines, vector‑based RAG models,...
How AI Covered a Human’s Paternity Leave // Quinten Rosseel
During a head of data’s paternity leave, a logistics SaaS firm relied on an AI analyst named “Wobby” to handle incoming data questions. The agent answered roughly 60 % of queries, demonstrating that a well‑engineered AI can fill staffing gaps without...
MCP Security: The Exploit Playbook (And How to Stop Them)
The video spotlights the rapid rise of the MCP (Model‑Centered Programming) standard since its November 2024 launch and the stark security lag that now threatens its expanding ecosystem. While major platforms are racing to support MCP, developers are left scrambling to...
The Future of Coding: AI Agents & the Next Tech Revolution // Ricky Doar
The conversation centers on Cursor, an AI‑driven coding assistant, and how developers are adapting to a new paradigm where large language models act as pair programmers or autonomous agents. Ricky Doar and his guest discuss the rapid adoption of Cursor...
Fast & Asynchronous: Drift Your AI, Not Your GPU Bill // Artem Yushkovskiy
The talk introduced ASEA, an open‑source asynchronous‑actor framework designed to replace traditional batch pipelines for generative AI workloads. By decoupling each processing step into self‑hosted GPU actors that communicate via message queues, the team at a global food‑delivery platform eliminated...
Beyond the Gold Standard: Evaluating and Trusting Agents in the Wild // Sanjana Sharma
AI agents look impressive in demos, but production reliability hinges on context, evaluation, and trust. Sanjana Sharma argues enterprises must shift from model‑first to system‑first thinking, embedding explicit business rules, subject‑matter‑expert (SME) heuristics, and versioned context layers. The talk outlines three...
Speed and Scale: How Today's AI Datacenters Are Operating Through Hypergrowth
The video, hosted by Chris, co‑founder and CEO of Netbox Labs, examines the unprecedented speed and scale of today’s AI datacenter construction. He frames Netbox as the de‑facto system‑of‑record that tracks everything from power and cooling to rack‑level configurations, giving...
