# agentic-engineering Published articles for agentic-engineering. This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles. ## How to Decide How Much Authorship AI Should Have in Software Work DevFeed: [How to Decide How Much Authorship AI Should Have in Software Work]() Original publisher: [Read original article]() Author: Kyle Humphrey Published: 2026-09-17T12:00:32Z Content type: opinion Language: en Sources: [Atomic Object]() Topics: [agentic-engineering](), [Artificial Intelligence](), [implementation](), [Software]() Tags: [agentic](), [agentic-engineering](), [ai](), [artificial-intelligence](), [development-practices](), [engineering](), [implementation](), [ownership](), [software](), [workflow]() ### AI overview This commentary examines how AI-generated meeting summaries, backlog items, and implementation work can introduce vocabulary, decisions, scope, and structure that a team does not recognize or own. It argues that the artifact's purpose should guide how much authorship is delegated to AI, using Curiosity, Craft, and Care as principles for making that decision. ### Source excerpt Atomic's Agentic Engineering for Teams describes how work moves through a product backlog, definition, planning, implementation, and review when agents do much of the building. While working with a small team during a Research, Design, and Planning (RDP) engagement, I started paying attention to a smaller decision inside that process: how much authorship we give [...] The post Use Curiosity, Craft, and Care to Decide What AI Should Write appeared first on Atomic Spin. ## Quoting Laurie Voss DevFeed: [Quoting Laurie Voss]() Original publisher: [Read original article]() Author: Simon Willison Published: 2026-09-14T14:34:29Z Content type: opinion Language: en Sources: [Simon Willison's Weblog]() Topics: [Code](), [Software]() Tags: [agentic](), [agentic-engineering](), [agentic-engineering-63](), [ai](), [ai-2-236](), [careers](), [careers-83](), [deep](), [deep-blue](), [deep-blue-12](), [generative](), [generative-ai](), [generative-ai-1-982](), [laurie-voss](), [laurie-voss-6](), [llms](), [llms-1-948]() ### AI overview A quotation from Laurie Voss argues that the cost of writing code has collapsed and that reviewing, fixing, and operating software may follow. It suggests that identifying user needs, defining them precisely, and making software pleasant to use could become the dominant remaining work as software production expands. ### Source excerpt The cost of writing code collapsed, and the cost of reviewing, fixing and operating it is following, and I'm assuming it gets there. What's left of making software is finding out what people actually want, defining it precisely, and making it pleasant to use. That cost is per piece of software and doesn't transfer, so as the amount of software goes to infinity, which it will because there's no ceiling on demand, that cost becomes the whole job. -- Laurie Voss, We are all Product Engineers now Tags: laurie-voss, generative-ai, agentic-engineering, ai, llms, deep-blue, careers ## Agentic Engineering in Python: From Vibes to Evidence DevFeed: [Agentic Engineering in Python: From Vibes to Evidence]() Original publisher: [Read original article]() Author: Ben Batman Published: 2026-09-14T14:00:00Z Content type: tutorial Language: en Sources: [Real Python]() Topics: [Loop Engineering](), [AI Agent](), [Python](), [Transformers](), [Code review](), [Test-driven development]() Tags: [agentic](), [agentic-engineering](), [ai-coding-agents](), [cheat-sheet](), [code-review](), [learning](), [python](), [tests]() ### AI overview This tutorial presents agentic engineering in Python as a disciplined workflow for delegating bounded tasks to AI coding agents while retaining human control over intent, constraints, review, and release decisions. It explains execution and acceptance loops, the Engineering Evidence Ladder, and the RECAP review method for evaluating agent-generated changes with tests, type hints, refactoring, and code review. ### Source excerpt Move from vibe coding to agentic engineering in Python, using tests, types, and code review to prove an AI agent's changes are safe to keep. ## Boris Cherny on guardrails for Claude-written production code DevFeed: [Boris Cherny on guardrails for Claude-written production code]() Original publisher: [Read original article]() Author: Simon Willison Published: 2026-09-11T17:47:11Z Content type: opinion Language: en Sources: [Simon Willison's Weblog]() Topics: [Claude](), [anthropic](), [AI-assisted coding](), [Security]() Tags: [agentic-engineering](), [agentic-engineering-63](), [ai](), [ai-2-236](), [ai-assisted-programming](), [ai-assisted-programming-407](), [anthropic](), [anthropic-336](), [boris-cherny](), [boris-cherny-4](), [claude](), [claude-310](), [claude-code](), [claude-code-127](), [coding-agents](), [coding-agents-248](), [generative-ai](), [generative-ai-1-982](), [llms](), [llms-1-948](), [security]() ### AI overview Boris Cherny says production code written by Claude should meet a higher standard than human-written code. He describes Anthropic's guardrails, including lint rules, tests, Claude-driven end-to-end tests, daily Claude-powered fuzzing, automated code and security reviews, and automated refactoring. ### Source excerpt Production code written by Claude should have a higher bar than if it was written by a human. At Anthropic, we have many guardrails in place to make sure this is happening: lots of lint rules, lots of tests, Claude-driven end to end tests, Claude-powered fuzzers running daily, automated code reviews and security reviews, automated code refactoring, and so on. Without these, you can end up with a mess that is hard to maintain down the line. -- Boris Cherny Tags: claude, ai, claude-code, llms, coding-agents, ai-assisted-programming, generative-ai, agentic-engineering, boris-cherny, anthropic ## Datasette 1.0a39 and 0.65.4 security releases DevFeed: [Datasette 1.0a39 and 0.65.4 security releases]() Original publisher: [Read original article]() Author: Simon Willison Published: 2026-09-11T03:27:16Z Content type: release Language: en Sources: [Simon Willison's Weblog]() Topics: [releases](), [Security](), [audit](), [AI-assisted coding](), [ai security](), [agentic-engineering](), [Generative AI](), [Large Language Model](), [Claude](), [gpt-6-astra]() Tags: [agentic-engineering](), [agentic-engineering-63](), [ai](), [ai-2-236](), [ai-security-research](), [ai-security-research-42](), [coding-agents](), [datasette](), [datasette-1-542](), [generative-ai](), [generative-ai-1-982](), [llms](), [llms-1-948](), [releases](), [releases-31](), [security](), [security-634](), [tests]() ### AI overview Datasette 1.0a39 and 0.65.4 are security patch releases for the alpha and stable series. They address issues found during an extensive audit, including instances that mix public and private tables. The audit used Claude Fable 5.1, GPT-5.6, and GPT-6 Astra alongside human review and automated tests. ### Source excerpt Datasette 1.0a39 and 0.65.4 security releases Today we're releasing two new security patch versions of Datasette: 1.0a39 and 0.65.4 - one for the current alpha series and one for the stable 0.65.x family. These are security fixes which you should apply if you are running a Datasette instance on the public web - in particular if that instance mixes both public and private tables. Following issues reported by Sevban Dönmez, Alex Garcia and I ran an extensive audit of Datasette using Claude Fable 5.1, GPT-5.6, and GPT-6 Astra. We then spent almost a week collaborating on and reviewing the fixes. They helped find some very subtle bugs. We'll be incorporating security audits by frontier models into all of our development work going forward. Alex came up with a way of splitting the work which I found extremely productive: Alex Garcia and I worked together running and then responding to the audit, working in a shared private repository. For most of the issues we split the work: one of us would create the automated tests highlighting the issue, then the other would implement the fix. This ensured that two separate humans had eyes on each of the issues, in addition to our coding agents running different models. Tags: releases, security, ai, datasette, generative-ai, llms, agentic-engineering, ai-security-research ## Coding Challenge #134 - Agentic Engineering Graph DevFeed: [Coding Challenge #134 - Agentic Engineering Graph]() Original publisher: [Read original article]() Author: John Crickett Published: 2026-08-29T08:01:12Z Content type: tutorial Language: en Sources: [Coding Challenges]() Topics: [agentic-coding](), [Orchestration](), [coding](), [MCP Server](), [Langgraph](), [MCP](), [Chaos Engineering](), [ide]() Tags: [agent](), [agentic](), [agentic-coding](), [agentic-engineering](), [ci](), [code](), [distributed-systems](), [langgraph](), [news](), [orchestration](), [review](), [sonarqube](), [systems]() ### AI overview A coding challenge guides readers through building an agentic engineering graph: a small orchestration system that runs an autonomous coding loop of planning, coding, testing, review, and retry. It uses scripts and files for dispatch and persisted state, adds a deterministic Sonar static-analysis quality gate, and then replaces it with an AI-augmented review node using the SonarQube MCP Server. ### Source excerpt This challenge is to build your own agentic engineering graph. ## Agentic Engineering 101 DevFeed: [Agentic Engineering 101]() Original publisher: [Read original article]() Author: Daniel Moka Published: 2026-08-12T05:01:48Z Content type: tutorial Language: en Sources: [Craft Better Software]() Topics: [Loop Engineering](), [Prompt Engineering](), [context](), [Artificial Intelligence]() Tags: [agent](), [agentic](), [agentic-engineering](), [agents](), [ai](), [context-engineering](), [engineering](), [prompt](), [prompt-engineering]() ### AI overview This tutorial introduces agentic engineering as a set of layers around an AI model. It describes prompt engineering, context engineering, harness engineering, loop engineering, graph engineering, and memory engineering, with detailed guidance in the supplied text on prompt design and context curation. ### Source excerpt Prompt vs Context vs Harness vs Loop vs Graph Engineering ## C31. AI-Native Practitioner PM DevFeed: [C31. AI-Native Practitioner PM]() Original publisher: [Read original article]() Author: Paweł Huryn Published: 2026-07-15T20:22:00Z Content type: article Language: en Sources: [The Product Compass]() Topics: [Artificial Intelligence](), [Claude](), [Claude Code](), [codex](), [agentic-engineering](), [Large Language Model]() Tags: [agentic](), [agentic-engineering](), [agents](), [ai](), [claude](), [claude-code](), [codex](), [guides]() ### AI overview This article introduces the C31 AI-Native Practitioner PM module and explains how it follows the C30 Claude and Codex Certified PM program. It outlines the certification path, including a knowledge test for C30 and planned exercises covering knowledge bases, agent memory, product discovery, prototyping, AI evaluations, and agentic engineering for C31. ### Source excerpt A complete path: live sessions, guides, , support, and your digital AI PM credentials. ## Owning Accountability for Agentic Engineering Systems DevFeed: [Owning Accountability for Agentic Engineering Systems]() Original publisher: [Read original article]() Author: Addy Osmani Published: 2026-07-15T00:00:00Z Content type: opinion Language: en Sources: [Addy Osmani]() Topics: [Artificial Intelligence](), [trust](), [implementation](), [coding]() Tags: [agentic-engineering](), [agents](), [delegation](), [quality](), [trust](), [verification]() ### AI overview This commentary argues that engineers must own the accountability boundary around agentic engineering systems. It defines quality as pre-release checks, verdict as the human production decision, and answerability as the ability to explain why a decision was made. It presents agents as models combined with harnesses and repeatable investigate-implement-verify cycles, with humans retaining responsibility for decisions at the boundary. ### Source excerpt Engineers need to own the outer loop, the accountability for these systems. A written version of my AI Engineer World's Fair 2026 closing keynote: quality, verdict, and answerability, the three hidden costs of delegation, and building software factories you can stand behind. ## 19 Hands-On Video Guides for PMs DevFeed: [19 Hands-On Video Guides for PMs]() Original publisher: [Read original article]() Author: Paweł Huryn Published: 2026-07-14T14:51:34Z Content type: article Language: en Sources: [The Product Compass]() Topics: [Learning](), [Claude Code](), [codex](), [agentic-engineering](), [vs-code](), [Extension](), [Command-line interface](), [n8n](), [Multi Agent Systems]() Tags: [agentic-engineering](), [claude-code](), [cli](), [codex](), [cursor](), [engineering](), [extension](), [guides](), [learning](), [multi-agent-systems](), [n8n](), [video]() ### AI overview The article announces an Events Archive containing 19 recordings of AI product-management sessions, including demonstrations, question-and-answer segments, timestamps, free templates, and linked resources. It highlights three free sessions on Claude Cowork, Codex App, and Claude Code, and describes upcoming sessions and an AI PM Learning Program. It also reports the author's Grok Build for VS Code extension reaching more than 20,000 installs and about 1,400 daily active users. ### Source excerpt The Events Archive is live: Claude Code, Cowork, Codex, n8n, Lovable, and more. Start with the 3 free ones. ## Why loop engineering needs a human at the boundary DevFeed: [Why loop engineering needs a human at the boundary]() Original publisher: [Read original article]() Author: Addy Osmani Published: 2026-07-09T14:31:24Z Content type: opinion Language: en Sources: [Elevate]() Topics: [Loop Engineering](), [Software](), [systems]() Tags: [agentic-engineering](), [agents](), [engineering](), [observability](), [process](), [software](), [systems](), [trust](), [verification]() ### AI overview The article argues that engineers must own the outer loop of agentic software systems. It defines quality checks, production verdicts, and answerability as human responsibilities, while describing agents as models combined with tools, memory, permissions, sandboxes, tests, observability, and recovery. ### Source excerpt Why loop engineering needs a human at the boundary ## Agentic Autonomy Levels DevFeed: [Agentic Autonomy Levels]() Original publisher: [Read original article]() Author: Addy Osmani Published: 2026-07-03T14:30:33Z Content type: article Language: en Sources: [Elevate]() Topics: [Artificial Intelligence](), [Orchestration](), [Agent Skill](), [Claude](), [Claude Code](), [codex]() Tags: [agent](), [agentic](), [agentic-engineering](), [agents](), [ai](), [claude](), [claude-code](), [codex](), [delegation](), [multi-agent](), [orchestration](), [parallel](), [scopes](), [trust](), [verification]() ### AI overview This article proposes a two-axis model for evaluating agentic engineering: agency, describing how independently a single agent operates, and orchestration, describing the skill of coordinating multiple agents. It argues that a single autonomy ladder is insufficient for multi-agent work and emphasizes matching autonomy levels with appropriate verification. ### Source excerpt A working model of autonomy for agentic engineering ## Agentic Autonomy Levels DevFeed: [Agentic Autonomy Levels]() Original publisher: [Read original article]() Author: Addy Osmani Published: 2026-07-02T00:00:00Z Content type: article Language: en Sources: [Addy Osmani]() Topics: [Artificial Intelligence](), [Orchestration](), [Agent Skill]() Tags: [agentic](), [agentic-engineering](), [ai](), [delegation](), [multi-agent](), [orchestration](), [scale](), [scopes](), [trust](), [verification]() ### AI overview The article argues that agent autonomy should be evaluated on two separate axes: agency, describing how independently an agent pursues a goal, and orchestration, describing the ability to coordinate multiple agents. It emphasizes choosing an autonomy level per task and defending that choice through verification. ### Source excerpt The action in agentic engineering has moved from prompting to operating. Autonomy isn't one ladder, it's two axes (agency and orchestration) and six levels you move between per task. The real question is what level a task deserves, and what verification makes that level defensible. ## Vibe once, run anywhere with Antigravity and Flutter DevFeed: [Vibe once, run anywhere with Antigravity and Flutter]() Original publisher: [Read original article]() Author: Craig Labenz Published: 2026-06-29T16:01:02Z Content type: article Language: en Sources: [Flutter - Medium]() Topics: [Flutter](), [AI-assisted coding](), [cross-platform](), [Artificial Intelligence](), [Dart](), [Large Language Model]() Tags: [agentic](), [agentic-engineering](), [ai](), [antigravity](), [code](), [coding-agents](), [cross-platform](), [dart](), [flutter](), [flutter-app-development](), [game-development](), [google-antigravity]() ### AI overview This article describes how a team used Google's Antigravity and Flutter to build and launch DashLander, a moonlander-style game set on procedurally generated asteroids. It explains how coding agents handled assets, code, testing, marketing, and deployment, and why Flutter's cross-platform model and Dart's strong typing can support agentic development. ### Source excerpt co-authored with Rody Davis Coding agents and the way we use them has already dramatically evolved since their arrival just a handful of months ago. Initially, the focus was heavily on observation -- reviewing every single output line by line. But as models rapidly increased in capability, the industry shifted toward true agentic engineering. Today, developers, PMs, and designers are wearing multiple hats, focusing on lofty conceptual goals and letting agents tackle the individual components. To explore this new frontier, our team wanted to build an experience that showcased this exact workflow. We wanted to build a game, generate its assets, write the marketing page, and deploy the whole thing, all using Google's premiere AI-native platform: Antigravity. Antigravity brings the best of Google into one place, using a tight feedback loop of planning, acting, and verifying. It creates artifacts, writes code, runs tests, and even clicks buttons in your UI to ensure it actually completed the task correctly. The result of our agentic adventure is DashLander -- a moonlander-style game set on procedurally generated asteroids. Here is the story of how we built it. Why Flutter in the AI era? A lot of people might be thinking: If agents can write native code, why not just have them write Android, iOS, and web apps entirely separately? That is a valid question; and is what originally inspired the concept of Flutter as a "Vibe once, run anywhere" UI toolkit. Having a single source of truth is just as important for AI as it is for humans. By having an agent write a single, cross-platform app, teams can eliminate the subtle bugs that inevitably creep in between different languages or platform-specific paradigms. Furthermore, Dart's strong typing gives LLMs excellent feedback. Languages with looser typing systems require dramatically more analysis by the LLM to know whether a given piece of code is correct across all scenarios. Flutter and Dart, on the other hand, use their analysis ## Cosmos is a shared platform for coordinating AI agents across software development workflows DevFeed: [Cosmos is a shared platform for coordinating AI agents across software development workflows]() Original publisher: [Read original article]() Author: Jeff Morhous Published: 2026-06-24T10:16:51Z Content type: opinion Language: en Sources: [The AI-Augmented Engineer]() Topics: [Cosmos](), [agentic-engineering](), [AI Development](), [AI-assisted coding](), [Software Engineering]() Tags: [agent](), [agentic](), [agentic-engineering](), [ai](), [claude-code](), [engineering](), [generation](), [software-development](), [workflow]() ### AI overview The article argues that AI agents are moving software development beyond the IDE and that teams need an operating model for coordinating agent-driven work. It presents Cosmos, Augment's unified cloud agents platform, as a shared system for workflows including triage, specification, implementation, review, testing, deployment, and feedback. ### Source excerpt Cosmos is a new tool for AI-native engineering teams ## What Product Managers Review in Agentic Engineering DevFeed: [What Product Managers Review in Agentic Engineering]() Original publisher: [Read original article]() Author: Paweł Huryn Published: 2026-05-31T22:50:22Z Content type: opinion Language: en Sources: [The Product Compass]() Topics: [agentic-engineering](), [prompt](), [Job]() Tags: [agentic](), [agentic-engineering](), [engineering](), [prompt]() ### AI overview The article argues that product managers working with agentic engineering do not need to code, but should review the resulting artifacts. It also presents this as an evolving product-management responsibility and mentions a prompt pack. ### Source excerpt You don't have to code. What you review instead, why it's the PM job now, and the prompt pack for agentic engineering. ## Designing a team of agents DevFeed: [Designing a team of agents]() Original publisher: [Read original article]() Author: Nicolas Fränkel Published: 2026-05-03T00:00:00Z Content type: opinion Language: en Sources: [Nicolas Fränkel]() Topics: [Artificial Intelligence](), [Claude Code](), [Software Engineering](), [Code]() Tags: [agentic-engineering](), [agents](), [ai](), [autonomous-agents](), [claude-code](), [software-engineering](), [technical]() ### AI overview The author describes experimenting with AI for software engineering by designing a team of autonomous agents to implement features from design through implementation. The article discusses coding assistants, agentic engineering levels, Claude Code's experimental Agent Teams feature, and subagents. ### Source excerpt I continue to experiment with AI in the context of software engineering. I'm fortunate that my team supports me in exploring different ways to improve our daily work. This week, I designed a team of autonomous agents to implement features, from design to implementation. Why autonomous agents? A long time ago, we were delighted when the IDE offered auto-completion. In the previous two years, things have changed. A lot. Coding assistants have become our primary interfaces for coding. ## The Multi-Model Playbook DevFeed: [The Multi-Model Playbook]() Original publisher: [Read original article]() Author: Sumedh Bhattacharya Published: 2026-01-16T16:02:06Z Content type: tutorial Language: en Sources: [Benchling]() Topics: [Large Language Model](), [ai-coding](), [Structured-data](), [agent observability](), [pdf]() Tags: [agentic-engineering](), [ai](), [ai-agents](), [analysis](), [benchling](), [biotechnology](), [caching](), [code](), [coding](), [comparisons](), [llm](), [model](), [parallel](), [patterns]() ### AI overview Benchling describes patterns for building production AI systems across multiple model providers. The article covers modular task decomposition, prompt structure, caching, structured data presentation, provider comparisons, and applying these principles to AI coding assistants. It reports using OpenAI GPT, Anthropic Claude, Google Gemini, Meta Llama, and Amazon Nova in agent systems such as the Data Entry Agent and Compose. ### Source excerpt The Multi-Model Playbook: Patterns in Agentic Engineering Building production AI systems that work reliably across multiple model providers requires more than just swapping API keys. Over the past year, working on AI agents like the Data Entry Agent and Compose Agent at Benchling, I've learned that successful multi-provider strategies come down to understanding what's universal versus what's provider-specific, and designing around those constraints. The clearest revelation here was that the architectural principles underlying reliable software -- modularity, separation of concerns, clear interfaces -- apply just as fundamentally to AI systems as they do to traditional code. The Data Entry Agent (DEA) extracts structured data from PDFs and images, while Compose is an agent that helps scientists write electronic lab notebooks (ELNs) by extracting content from attached files, connecting that with data in Benchling's Registry, and outputting structured scientific protocols, analysis, and more. These systems currently support five different model families (OpenAI GPT, Anthropic Claude, Google Gemini, Meta Llama, and Amazon Nova), typically using four in any given run. This experience has revealed patterns that hold true across providers -- patterns around task decomposition, prompt structure, caching strategies, and data presentation. While each provider has its quirks, these foundational strategies have proven consistently effective. In this post, I'll cover: How to break down problems for optimal LLM performance Why the system versus user prompt distinction matters for caching Best practices for presenting structured data as context Practical comparisons between model providers How to apply these principles when using AI coding assistants. Breaking Down Problems: Small & Complex versus Large & Simple LLMs lose accuracy when handling multiple separate tasks simultaneously or when operating on large input contexts. The sweet spot is to give them either a small, complex task ## AI Coding Often Handles the First 70% but Leaves Difficult Edge Cases and Integration Work DevFeed: [AI Coding Often Handles the First 70% but Leaves Difficult Edge Cases and Integration Work]() Original publisher: [Read original article]() Author: Franciska Dethlefsen Published: 2025-11-06T00:00:00Z Content type: opinion Language: en Sources: [Zed Industries - Blog]() Topics: [Artificial Intelligence](), [ai-coding](), [code productivity](), [Developer experience](), [Google]() Tags: [agentic-engineering](), [ai](), [ai-coding](), [developer](), [developer-experience](), [google](), [productivity](), [trust]() ### AI overview A discussion with Addy Osmani examines the limits of AI coding. AI tools can quickly produce scaffolding and obvious patterns, but edge cases, security, debugging, and production integration remain difficult. The article also describes rising AI-code adoption alongside declining trust in generated code. ### Source excerpt We hosted Addy Osmani, who works on AI and dev tools at Google's Chrome Developer Experience team, to talk about what he calls the "70% problem" in AI coding. Over the past two years, Addy has been tracking AI adoption patterns at Google, where over 30% of code is now AI-generated, and across the industry at conferences like Lead Dev. He's studied trust metrics, productivity claims, and what actually happens when teams scale agentic engineering beyond prototypes. ## Mitchell Hashimoto's Approach to Using AI When Building Ghostty DevFeed: [Mitchell Hashimoto's Approach to Using AI When Building Ghostty]() Original publisher: [Read original article]() Author: Franciska Dethlefsen Published: 2025-06-19T00:00:00Z Content type: article Language: en Sources: [Zed Industries - Blog]() Topics: [Artificial Intelligence](), [Code](), [Large Language Model](), [Refactoring](), [Testing](), [Tooling]() Tags: [agentic](), [agentic-engineering](), [ai](), [claude](), [llms](), [refactoring](), [testing](), [tooling]() ### AI overview Mitchell Hashimoto describes using AI while building Ghostty, treating it as a tool for scoped debugging, refactoring, and cleanup while he retains responsibility for architecture and quality. He also discusses working alongside AI-assisted tools during testing and stress testing. ### Source excerpt Mitchell Hashimoto walked Richard Feldman through his approach to using AI when building Ghostty.