Resources
The reading list behind the job board.
I built this board by reading everything I could find on how companies actually deploy AI coding agents. These are the articles, case studies, and X threads that shaped my thinking — all published since late 2025, when this space started moving for real.
March 2026
Hiring agentic engineers?
Before you dive into the reading list, I've put together a practical Interview Guide — 9 questions with green and red flags specifically designed for evaluating agentic maturity. It works for non-technical interviewers too.
Ready to reach this talent pool? Post a job — it takes five minutes.
The Shift
The pieces that named and framed what's happening. Start here if you need to explain agentic engineering to someone — or understand it yourself.
Agentic Engineering Andrej Karpathy · X · Feb 2026
The post that gave the discipline its name. On the one-year anniversary of his “vibe coding” tweet, Karpathy drew the line: agentic engineering is orchestrating agents who write the code, not writing it yourself.
The Third Era of AI Software Development Michael Truell, Cursor CEO · Feb 2026
Tab autocomplete → synchronous agents → autonomous cloud agents. Truell maps the trajectory with hard numbers: 35% of Cursor's internal PRs are now agent-written, and agent users outnumber tab users two-to-one.
Harness Engineering: Leveraging Codex in an Agent-First World OpenAI · Feb 2026
Three engineers, one million lines of production code, zero handwritten. OpenAI coined “harness engineering” — the practice of designing the constraints, context, and verification loops that make agents productive.
Exploring Generative AI: Harness Engineering Birgitta Böckeler · Martin Fowler · Feb 2026
The clearest breakdown I've found of what harness engineering means in practice — three pillars (Context Engineering, Architectural Constraints, Garbage Collection) with a critical feedback loop insight. This is the blueprint.
How Companies Do It
Production deployments from companies you know by name. These aren't pilots — these are systems running at scale, and the numbers are worth reading carefully.
Minions: Stripe's One-Shot End-to-End Coding Agents Stripe Engineering · Feb 2026
Stripe merges 1,300 agent-written PRs every week. Their key finding — constraints, not model intelligence, drive reliable output — is the single most important engineering insight in this space.
How AI Is Transforming Work at Anthropic Anthropic · Dec 2025
Internal data from the company building the model. Some engineers produce 100% of their code through AI — this is the benchmark for what full agentic adoption actually looks like from the inside.
Building a C Compiler with a Team of Parallel Claudes Anthropic Engineering · Feb 2026
Sixteen agents, 100,000 lines of compiler code, $20,000 in compute. A deep technical case study that shows what's possible when you build the right harness around genuinely hard problems.
Eight Trends Defining How Software Gets Built in 2026 Anthropic · Jan 2026
Anthropic's industry report with real enterprise case studies from Rakuten, TELUS, and Zapier. The central finding: developers integrate AI into 60% of their work while maintaining active oversight on 80-100% of delegated tasks.
How to Get Good
Practical guides for engineers who want to get better at working with agents, not just faster. These are the people actually building this way every day.
Real-World Feature Build with Claude Code: Every Step Explained Matt Pocock · AI Hero · Mar 2026
The most practical walkthrough I've seen of an end-to-end agentic workflow — from idea to shipped feature using Claude Code. Pocock shows every step, every prompt, every decision point. No theory, just craft.
5 Agent Skills I Use Every Day Matt Pocock · AI Hero · Mar 2026
Skills are reusable instruction sets that make agents consistent. Pocock — who pivoted from teaching TypeScript to teaching AI engineering — shares the five he reaches for daily and why they matter.
Towards Self-Driving Codebases Cursor · Feb 2026
The long-horizon vision: codebases designed from the ground up for autonomous agents to navigate. Worth reading now to understand what your architecture decisions today are actually setting up.
Scaling Long-Running Autonomous Coding Cursor · Jan 2026
What happens when you move from individual agent use to hundreds of concurrent agents on a single codebase. Key finding: hierarchical agent structures outperform flat ones, and prompts matter more than infrastructure.
7 Shifts the 99.99% Haven't Made Yet Tibo · X · Feb 2026
The behavioral changes that separate advanced agentic practitioners from casual AI users — from behavioral scenario engineering to outcome evaluation over code review. A sharp X thread that's worth bookmarking.
Where to Find Work
The places worth your attention if you're actively looking for agentic engineering roles or want to stay current in this space.
- Job boards: Agentic Engineering Jobs — the only board that filters by the specific AI tools a role requires (Cursor, Claude Code, Copilot, Codex), with agentic maturity levels so you know what you're actually walking into.
- Communities: Latent Space Discord (researchers and practitioners talking about what's actually happening in AI), Cursor Discord (the fastest way to learn workflow patterns from power users), Claude Developers Discord (Claude Code tips and agent development patterns).
- Newsletters: Latent Space (the paper-to-practice newsletter for AI engineers), Pragmatic Engineer (strong signal-to-noise on engineering culture and tooling shifts), TLDR AI (daily digest of the headlines worth knowing about).
- People to follow on X: @karpathy, @mattpocockuk, @swyx, @mckaywrigley, @tibo_maker
Know a resource that belongs here? Send it to hello@maxim-buz.dev.
Looking for agentic engineering roles? Browse jobs →