the same objective: shorten feedback loops while making delivery safer and more repeatable. 2006–09 CLOUD + DEVOPS Programmable infrastructure and DevOpsDays challenge the wall between development and operations. 2010–16 AUTOMATION + SRE CI/CD, containers and Kubernetes scale delivery; Google formalizes reliability engineering and toil reduction. LATE 2010s PLATFORM ENGINEERING Cloud-native systems raise cognitive load. Platform teams and Internal Developer Platforms emerge. 2020s AI-ASSISTED DEVELOPMENT Generative and agentic AI accelerate the loop—but observability, controls and reliability remain essential. Evolution, not replacement: new disciplines operationalize and scale the same collaborative principles. Source: Yurie Mori, SRE Magazine, 1 Aug 2025 · sre-magazine.net/articles/9/yuriemori/
article shifts the question from “What tools do we have?” to “Have we achieved the Three Ways?” 01 FLOW Move value left → right Small batches · less WIP Fewer handoffs and delays Quality built into delivery 02 FEEDBACK Learn right → left Problems visible early Production and customer signals Fast, shared problem-solving 03 CONTINUAL LEARNING + EXPERIMENTATION Improve the system Safe experiments Knowledge shared across teams Daily improvement and resilience THE UNFINISHED DEVOPS QUESTION Have we built these organizational capabilities—or only upgraded the toolchain? Technology is evidence of progress—not proof that the Three Ways are complete. Source: Yurie Mori, SRE Magazine, 1 Aug 2025 · sre-magazine.net/articles/9/yuriemori/
them as complementary layers—not competing job titles or tool categories. DEVOPS Process, Culture, Technology Shared ownership Fast feedback Continuous improvement Sets the intent and culture SRE Reliability engineering Software engineering for operations Measurable reliability Automation and less toil Implements DevOps for production PLATFORM ENGINEERING Productized enablement Internal Developer Platform Self-service “paved roads” Governance with autonomy Scales the developer experience DevOps provides the why · SRE makes reliability explicit · Platform Engineering makes good practice consumable at scale Source: Yurie Mori, SRE Magazine, 1 Aug 2025 · sre-magazine.net/articles/9/yuriemori/
“DEVOPS-AS-A-SERVICE” An Internal Developer Platform turns repeatable DevOps practices into self-service workflows. It reduces cognitive load while keeping security, policy and operational guardrails visible. AGENTIC AI ACCELERATES THE LOOP AI can assist coding, testing, deployment and operations. Faster action increases—not decreases—the need for observability, reliable feedback, human judgment and organizational learning. SPEED RELIABILITY CONTINUOUS FEEDBACK The enduring challenge is not choosing one side—it is learning fast enough to improve both. Source: Yurie Mori, SRE Magazine, 1 Aug 2025 · sre-magazine.net/articles/9/yuriemori/
era removed a different boundary. AI makes intent the new point of separation. 01 AGILE User Dev Share customer value WHAT should we build? 02 DEVOPS Dev Ops Share delivery ownership HOW do we deliver safely? 03 AI-ASSISTED ENGINEERING Human AI Share context and intent WHY was this choice made? AI can generate the artifact. The harder problem is transferring the reasoning behind it.
moves intent—not only code—through the engineering system. 1 CAPTURE INTENT ADRs · design docs PR rationale · domain goals 2 CONNECT KNOWLEDGE Code · issues · runbooks SLOs · incidents · wikis 3 TRANSPORT CONTEXT MCP · RAG · knowledge graph Agent-to-agent exchange 4 AI ACTS Code · tests · docs Pull requests · operations 5 MEASURE & LEARN Validate outcomes and context Feed decisions back into knowledge LOOP BACK Learning updates the context for the next decision. These technologies do not merely make AI smarter—they deliver the context required for trustworthy action.
context; culture determines whether useful context exists in the first place. WRITE FOR FUTURE HUMANS + AI 1 Record architecture decisions 2 Explain trade-offs in pull requests 3 Maintain READMEs and design docs 4 Make reviews and discussions searchable 5 Use InnerSource to keep knowledge open CALMS—REINTERPRETED FOR AI-ASSISTED ENGINEERING C Culture Preserve context as a team habit A Automation Include AI agents in the delivery system L Lean Reduce human cognitive load through collaboration M Measurement Measure context sufficiency—not output volume S Sharing Move from knowledge sharing to context sharing The next DevOps frontier is to overcome silo; Human-Agents, Agents-Agents, and Human-Human(still matters)
goes on.” Principles endure Collaboration, shared ownership and continuous improvement remain the foundation. Implementations evolve SRE, platforms and AI reshape how those principles are delivered. Balance is continuous Speed and reliability improve through evidence, feedback and learning. Source: Yurie Mori, SRE Magazine, 1 Aug 2025 · sre-magazine.net/articles/9/yuriemori/
Engineering: A Journey Toward the Ideal Development Environment Explore the history, relationships and next evolution of modern software delivery. Yurie Mori · SRE Magazine · 1 Aug 2025 article-qr.png SCAN TO READ sre-magazine.net/articles/9/yuriemori/ sre-magazine.net/articles/9/yuriemori/ Full article available in Japanese
yuriemori
diggymo
oracle4engineer
ks91
shin1988
miyukichi_ospf
maroon8021
kakehashi
ymty
kobayashiyorimitsu
vtryo
nomizone
ekispert_api
rabernat
badams
sarafernandez
inesmontani
akademiya2063
hatefulcrawdad
vanstee
aki_iinuma
andyhume
bkeepers
keithpitt
samanthasiow
