41 Curated Talks on Agentic Engineering
Below is curated set of 41 talks on agentic engineering, spanning foundational concepts, architecture patterns, the verification problem, context and reliability, and the GTM/business angle. Roughly ordered foundational to advanced within each section.
Foundational / Conceptual
Andrej Karpathy: From Vibe Coding to Agentic Engineering w/ Stephanie Zhan · Sequoia Capital · 2026-04-29 · 29:49 · The flagship talk. “Vibe coding raised the floor; agentic engineering raises the ceiling.” Software 3.0, LLMs as a new compute substrate. Start here.
Anthropic’s Boris Cherny: Why Coding Is Solved, and What Comes Next · Sequoia Capital · 2026-05-04 · 24:36 · The Claude Code lead on what remains once code generation is cheap. Same stage and same season as the Karpathy talk, and it answers the question Karpathy leaves open.
How We Build Effective Agents: Barry Zhang, Anthropic · AI Engineer · 2025-04-04 · 15:09 · The author of Anthropic’s “Building Effective Agents” essay distills it live. Don’t-build-an-agent-when-a-workflow-will-do pragmatism.
Don’t Build Agents, Build Skills Instead – Barry Zhang & Mahesh Murag, Anthropic · AI Engineer · 2025-12-08 · 16:22 · Zhang revising his own agents-versus-workflows thesis eight months later. Watch it directly after the entry above.
Building more effective AI agents · Anthropic · 2025-10-17 · 18:58 · Alex Albert with a multi-agent research co-author on what they learned building Claude’s agents.
Loop Engineering from First Principles — Kyle Mistele, HumanLayer · AI Engineer · 2026-07-25 · 17:57 · Derives the agent loop from scratch instead of inheriting a framework’s defaults. The best conceptual treatment of scaffold design in the set.
The Engineering of AI Agents: Context, Harnessing, and Autonomy · InfoQ · 2026-05-06 · 42:01 · Senior-engineer-oriented breakdown of the agent harness.
Code with Claude 2026: Opening Keynote · Claude · 2026-05-06 · 47:29 · Ami Vora, Cat Wu, and Boris Cherny on the move from line-level completion to goal-level delegation. Vendor keynote, and also the primary source for the position everything else is arguing with.
Architecture & Patterns
The Great Loops Debate — Dex Horthy, Geoff Huntley, Ian Livingstone, Greg Pstrucha, @insecure-agents · AI Engineer · 2026-07-17 · 60:16 · Five harness builders disagreeing on control-loop architecture in public. The single-threaded versus multi-agent question staged as an argument nobody wins, which is rarer and more useful than another pattern catalog.
Your Agent Didn’t Fail. Your Harness Did. — Vinoth Govindarajan, OpenAI · AI Engineer · 2026-07-29 · 18:25 · Control planes, session-state rehydration, single-writer execution, approval paths, auditability. The “push correctness into the runtime” argument made from OpenAI’s side of the table, in their vocabulary.
From Chaos to Choreography: Multi-Agent Orchestration Patterns That Actually Work — Sandipan Bhaumik · AI Engineer · 2026-04-08 · 26:28 · Treats multi-agent coordination as a distributed systems problem and names the specifics: a stale-cache race condition in a financial services deployment, immutable state snapshots, circuit breakers. Locally correct, globally wrong, with a named instance attached.
From Stateless Nightmares to Durable Agents — Samuel Colvin, Pydantic · AI Engineer · 2025-11-24 · 22:12 · Durable execution for agents from the author of the Pydantic AI and Temporal integration.
Code Execution with MCP: Fix Tool Token Bloat (Adam Jones, Anthropic) · Agentic AI Foundation · 2025-12-12 · 34:19 · Code-execution-as-tool-use from the Anthropic engineer who built it. The only MCP entry on the list.
Agentic AI Patterns by Kevin Dubois · Devoxx · 2026-02-27 · 48:32 · Theoretical overview of agentic patterns from a major dev conference.
Agentic AI Patterns · JChampions Conference · 2026-01-23 · 62:05 · Longer-form companion treatment of the same pattern catalog. Overlaps the Devoxx talk above; pick one.
Effective Agentic AI Development with Specs, Skills and Old Good Design Patterns | WAWTech 2025 · WAWTech · 2026-04-27 · 49:57 · Argues that success comes from engineering structure rather than from prompt-fiddling.
Verification, Trust, and Code That’s Almost Right
Six talks from mid-2026 arguing that generation stopped being the constraint and verification became it. This is the section closest to Reboot’s own thesis.
Harness Engineering is not Enough: Why Software Factories Fail — Dex Horthy, HumanLayer · AI Engineer · 2026-07-23 · 19:17 · Horthy ran an experiment from July to November 2025 where Opus-based agents shipped every stage with no human review. By November a single bug took weeks to find because the accumulated code was unmaintainable. Code that compiled, deployed, and broke later, with dates on it. Start this section here.
In the Land of AI Agents, the Verifiers Are King — Tariq Shaukat, Sonar · AI Engineer · 2026-07-20 · 18:53 · The argument that verification is now the binding constraint on agent-written code, from a CEO whose company sells static correctness analysis.
Understanding is the new bottleneck — Geoffrey Litt, Notion · AI Engineer · 2026-07-10 · 19:33 · The trust gap framed as a comprehension problem: humans can’t read enough of the output to know whether it’s right. Litt is a researcher rather than a vendor, so the talk sells nothing.
Simon Willison: Engineering practices that make coding agents work - The Pragmatic Summit · The Pragmatic Engineer · 2026-03-19 · 28:20 · The lethal trifecta, plus red-green TDD as the discipline that makes agent output checkable. Security and verification in one sitting.
Your Agent Failed in Prod. Good Luck Reproducing It. - Tisha Chawla & Susheem Koul, Microsoft · AI Engineer · 2026-06-28 · 14:09 · Nondeterminism as an operations problem. Reproducibility is the precondition for every reliability claim in the section below.
Agentic Security: Permissions, Provenance, and the Agent Supply Chain — Steve Yegge, Gas Town · AI Engineer · 2026-07-20 · 22:32 · Agent security as a provenance and supply-chain problem rather than a prompt-filtering one. Containment over filtering.
Context Engineering & Reliability (the hard part)
Advanced Context Engineering for Agents · YC Root Access · 2025-08-25 · 14:37 · Dex Horthy’s tight, high-signal summary of the field’s key insights. The short cut.
Context engineering with Dex Horthy · The Pragmatic Engineer · 2026-07-15 · 93:11 · The long cut, from the person who coined the term.
Context Engineering for Agents - Lance Martin, LangChain · Latent Space · 2025-09-11 · 63:18 · Deep dive on why agents fail in practice and how context management fixes it.
Context Engineering for AI Agents with LangChain and Manus · LangChain · 2025-10-14 · 60:53 · Production-grade context engineering with the Manus team.
Simplifying Context Engineering for AI Agents in Production · Temporal · 2025-10-22 · 62:07 · The durable execution angle, relevant to Temporal and workflow orchestration.
WTF Is the Context Layer? The Missing Infrastructure for Production Agents — Prukalpa Sankar · AI Engineer · 2026-07-14 · 20:53 · Context as infrastructure rather than prompt text. The substrate argument applied one layer up.
Anthropic Workshop: Build Agents That Run for Hours — Ash Prabaker & Andrew Wilson · AI Engineer · 2026-05-18 · 75:40 · Long-horizon agents, the frontier of reliability.
How to Build Reliable AI Agents (Context + Evals Explained) | Tobias Leong, Axium · Arize AI · 2026-04-23 · 47:47 · Demo-to-production journey with real eval practices.
How Lyft Builds Evals That Actually Matter in Production | Interrupt 26 · LangChain · 2026-06-15 · 17:11 · Eval design from a team with real traffic instead of a benchmark.
Opening Keynote: Engineering for Trust in the Era of Agentic AI · Database Trends and Applications · 2026-05-12 · 32:17 · Governance and trust as engineering concerns.
GTM / Business Angle
Measuring Exponential Trends Rising (in AI) — Joel Becker, METR · Latent Space · 2026-02-27 · 65:12 · The best long-form treatment of the METR randomized trial: why experienced developers were 19% slower while believing they were faster, and how time-horizon evals get misread. The number every investor conversation now cites.
The Current Impact of AI on Engineering Velocity: What 400 Companies Are Seeing · DX · 2026-04-27 · 26:15 · Cross-company measurement showing roughly 10 to 15% velocity gains. The strongest counterweight to vendor ROI claims, because it’s measurement across 400 companies.
Building world-class engineering teams in the age of AI - The Pragmatic Summit · The Pragmatic Engineer · 2026-02-24 · 33:36 · Thomas Dohmke (ex-GitHub CEO) and Rajeev Rajan (Atlassian CTO) on restructuring an engineering org around agents. The clearest buyer-side framing available.
Agentic AI and the Future of Enterprise Software in 2026 · CXOTalk · 2026-04-24 · 65:53 · Praveen Akkiraju of Insight Partners on ROI-based funding of agents, tracking tokens and tool-use against business outcomes, and which agent companies survive contact with enterprise buyers.
Bret Taylor of Sierra on AI agents, outcome-based pricing, and the OpenAI board · Stripe · 2026-03-10 · 101:42 · The outcome-based pricing argument from the operator who commercialized it.
Sequoia’s Alfred Lin: “Software Code Is No Longer a Moat” | 2026 Upfront Summit · Upfront Ventures · 2026-03-12 · 28:07 · What defensibility means when code generation is a commodity. Short and quotable for investor-facing work.
Panel: Agentic AI Debt: Stochastic Behavior & Change · Data Science Dojo · 2026-04-21 · 47:45 · Governance strategies and the hidden cost of stochastic systems, a sharp GTM and risk lens.
The Future of Agentic Workflows: A Panel Discussion with AI Experts · Auth0 · ~44:08 · Where agentic products are heading, plus identity and access implications. Embedding is disabled on this one, so link checkers report it dead; it’s live. Date unverified.
Enterprise Adoption is the Biggest Bottleneck for AI Agents | Randy Bias, Mirantis · TFiR · 2026-03-09 · 3:11 · Quick, punchy take on what’s actually blocking deployment.