Extreme Programming, AI and Modern Practices: XP in the Agentic Era

Extreme Programming (XP) represents one of the most radical departures from traditional software engineering paradigms. Positioning itself not merely as a set of technical instructions but as a holistic system of social change and technical rigor. At its core the methodology seeks to reconcile the inherent humanity of developers with the uncompromising demands of business productivity. Asserting that outstanding results are achieved only when these two forces are brought into harmony. The genesis of XP in the mid-1990s marked a transition from the “Big Design Up Front” (BDUF) era to an iterative and feedback-driven approach that celebrates human creativity while accepting human frailty. [1] As the discipline moves into the mid-2020s the integration of generative artificial intelligence and autonomous agents has initiated a “Software Engineering 3.0” era. Where XP’s core tenets of feedback, simplicity and continuous testing serve as the essential guardrails for a new paradigm of vibe coding and agentic orchestration. [4]

The Genesis of a Paradigm: The Chrysler Creation Story

The historical narrative of Extreme Programming is inextricably linked to the Chrysler Comprehensive Compensation (C3) payroll project which began in 1996. Kent Beck was originally brought in to consult on the performance of a failing Smalltalk system but upon discovering that the project lacked fundamental testing and stable requirements he proposed a radical restart. The project, which officially kicked off around Easter of 1996, served as the laboratory for what would become the XP methodology. The early C3 team included influential figures such as Ron Jeffries, who became the first full-time XP coach, and Martin Fowler, who assisted with analysis and testing frameworks. [1]

During the initial phase of the C3 project, Beck sought to “turn the dials to ten” on every practice known to produce value in software engineering. This meant that if testing was good, the team would test all the time; if simplicity was good, the team would build only the smallest possible solution to current problems; and if collaboration was good, the team would work in pairs and sit together in an open space. One of the most famous lessons from this period was the realization that “end-to-end is further than you think,” a phrase born out of a late-1996 crisis where the team had calculated numbers correctly but failed to export them in the required format for production. Despite such setbacks the project was a technical success producing a system that remained in production for three years and demonstrated that object technology could handle large-scale data processing with incredibly low defect rates. [1]

The Axiological Core: Values as the Source of Purpose

XP is built upon a hierarchy of values, principles and practices. Values provide the high-level criteria for judgment, representing the “roots” of why certain behaviors are preferred over others. Without these values, practices risk becoming rote activities performed for their own sake without a clear sense of direction. [1]

The foundational belief of XP is that software development is a person-to-person business. Most failures are attributed not to technical incompetence but to a lack of knowledge transfer or the presence of defensive social barriers. XP mandates that communication be direct and frequent. Favoring oral conversation over voluminous documentation. The system structure is communicated through tests and source code rather than static diagrams. Ensuring that the team shares a single living version of the truth. [1]

Simplicity and the Elimination of Over-Engineering

Simplicity is defined as the intellectual effort required to eliminate wasted complexity. XP asks developers to solve only today’s problem and resisting the instinct to build infrastructure for hypothetical future requirements. This value is context-dependent; a solution that is simple for a team experienced with parser generators may be complex for a team that is not. By maintaining the simplest possible design the team reduces the amount of code they must communicate about and maintain. Thereby increasing their agility in the face of change. [1]

Feedback Loops and the Shortening of Cycles

Change is inevitable in software development. Necessitating a strategy based on continuous adaptation rather than rigid adherence to a plan. Feedback in XP occurs at multiple timescales from the second-by-second interaction of pair programming to the minute-by-minute feedback of automated unit tests and finally to the weekly and quarterly reviews with customers. The methodology strives to shorten these cycles as much as possible, as the sooner a team knows they are headed in the wrong direction, the less it costs to correct course. [1]

Courage and Respect

Courage manifests as the willingness to tell the truth about progress even when it is unpleasant and the bravery to discard a design that no longer fits the system’s needs. It is a bias toward action that allows teams to move forward despite uncertainty. However, courage must be balanced by respect. Every person touched by the software project (programmers, sponsors, managers, and users) must be respected as a human being. This respect is the prerequisite for psychological safety which in turn enables effective communication and teamwork. [1]

The Intellectual Substrate: XP Principles

Principles bridge the gap between abstract values and situated practices. They provide the reasoning behind why a practice works and help teams improvise new practices when standard ones do not apply. [1]

Principle	Narrative Context and Application
Humanity	Software is written by people who have basic needs for safety, accomplishment, belonging, and growth. XP practices, such as energized work and limiting work hours, are chosen to respect these human limits. [1]
Economics	Development must be fiscally responsible. XP acknowledges the time value of money, aiming to earn revenue sooner and spend costs later through incremental design and pay-per-use deployment. [1]
Mutual Benefit	Solutions must benefit all stakeholders simultaneously. Automated tests benefit the current developer by clarifying design and the future maintainer by providing a safety net. [1]
Self-Similarity	Patterns that work at one scale should be applied to others. The “test-fail-pass” rhythm of coding is mirrored in the weekly cycle of release planning and the quarterly cycle of theme development. [1]
Improvement	“Perfect” is viewed as a verb rather than an adjective. The goal is not a perfect initial design but a process of continuous refinement in the light of experience. [1]
Diversity	Complex problems require multiple perspectives. Cross-functional teams (Whole Team practice) bring together diverse skills to identify pitfalls that a homogeneous group might miss. [1]
Reflection	Good teams analyze why they succeeded or failed. Reflection is woven into daily activities like pair programming and integration. Ensuring that learning is continuous and coupled with action. [1]
Flow	Value should be delivered in a steady stream rather than in large and risky chunks. This principle encourages small increments of value delivered frequently to maximize feedback. [1]
Opportunity	Problems are seen as chances for learning and system improvement. A person making a mistake is an opportunity for the team to adopt pair programming or improve automated testing. [1]
Redundancy	Critical problems should be addressed by multiple overlapping practices. Defects are mitigated by pairing, continuous integration and customer involvement. Ensuring that if one fails then others catch the error. [1]
Quality	Quality is not a control variable. Pushing quality higher (fewer defects and better design) actually results in faster and more predictable delivery over time. [1]
Baby Steps	Momentum is maintained through rapid small changes rather than risky momentous leaps. This is reflected in the test-first cycle and the frequency of integration. [1]
Accepted Responsibility	Responsibility cannot be assigned from above. It must be accepted by those performing the work. This ensures that authority and responsibility are aligned. [1]

Primary Practices: The Foundation of Engineering Excellence

Primary practices are those that any team can start with to see immediate improvements in effectiveness and humanity. These practices represent a vector toward the ideal state of development and providing clear goals for teams to strive toward. [1]

The Collaborative Workspace (Sit Together and Whole Team)

XP asserts that technical fixes alone are never enough to save a floundering project; the solution is almost always rooted in social dynamics. The practice of sitting together in an open space allows teams to communicate with all their senses. Picking up on the “hum” of productivity or the silence of risk. This is complemented by the Whole Team practice, which ensures that all skills necessary for success. Including those of database administrators, interaction designers, and architects are represented and identified primarily with the project team rather than their functional silos. The ideal team size is typically around twelve people though organizations can scale by fracturing larger problems into independent teams that integrate frequently. [1]

The Feedback Heartbeat (Stories and Planning Cycles)

Planning in XP is an ongoing activity rather than a phase. The team uses stories (units of customer-visible functionality) to plan work. Unlike “requirements,” stories carry early estimates that allow the business to make informed decisions about cost and value. The Weekly Cycle involves a meeting where progress is reviewed, new stories are selected and tasks are broken down and estimated by the developers who will implement them. This is nested within a Quarterly Cycle that focuses on larger themes and aligns the project with broader organizational goals. Slack is intentionally built into every plan. Providing minor tasks that can be dropped if the team falls behind. Thereby ensuring that commitments are met and trust is maintained. [1]

The Technical Inner Loop (TDD, Continuous Integration and Ten-Minute Build)

The technical rigor of XP is defined by a rapid inner loop of development. Test-Driven Development (TDD) requires developers to write a failing automated test before any production code. Ensuring that the system is always testable and that interfaces are decoupled from implementations. Continuous Integration mandates that these changes be merged and tested every few hours. Reducing the “divide and conquer” penalty that occurs when teams integrate code after weeks of isolation. Supporting this is the Ten-Minute Build, an ideal where the entire system can be compiled and all tests run in less than ten minutes, providing immediate feedback and reducing the stress of integration. [1]

Sustainable Design (Incremental Design and Energized Work)

Design in XP is a daily activity. Incremental Design suggests that the most effective time to invest in design is in the light of experience and deferring complexity until it is absolutely necessary. The most powerful design heuristic is “Once and Only Once,” where logic and structure exist in only one place to minimize the cost of future changes. This technical sustainability is mirrored in the human sustainability of Energized Work which insists that developers only work as many hours as they can remain productive. Long hours are viewed as a failure of planning and a risk factor for defects as insight comes most readily to a rested mind. [1]

Corollary Practices: Reaching Advanced Agility

Corollary practices are more difficult or dangerous to implement without first mastering the primary engineering and planning disciplines. They represent the “higher-level” capabilities of an Extreme Programming team. [1]

Real Customer Involvement: Truly visionary customers are made part of the team. Participating in planning and having a dedicated budget for the functionality they desire. [1]
Incremental Deployment: Teams avoid “big bang” deployments by taking over the workload of legacy systems early and in small stages which provide a safer path to modernization. [1]
Team Continuity and Shrinking Teams: Effective teams are kept together to preserve the value of their relationships. As a team becomes more capable its size is gradually reduced while its workload stays constant. Freeing up members to form new teams. [1]
Root-Cause Analysis: Every defect is an opportunity to improve the process. Teams write system-level tests for every bug and use the “Five Whys” to identify process or social failures. [1]
Shared Code and Single Code Base: Anyone on the team can improve any part of the system at any time. This is supported by a single code base where temporary branches live for no more than a few hours, preventing the waste of retrofitting fixes across multiple versions. [1]
Daily Deployment and Pay-Per-Use: Software is put into production every night, and revenue is realized through fees per transaction. This provides the ultimate feedback loop: direct financial data on the value of every story. [1]

Philosophical Roots: Beyond Taylorism

Extreme Programming is as much a social movement as a technical one, consciously rejecting the “Scientific Management” principles established by Frederick Taylor in the early 20th century. Taylorism assumes that workers are interchangeable cogs who must be told exactly what to do by elite engineers. A model that separates planning from execution and isolates quality into a separate department. This social structure is fundamentally unsuited to the creative and highly collaborative nature of software development. [1]

Instead, XP aligns with the philosophies of the Toyota Production System (TPS) and lean manufacturing. In TPS, the goal is to eliminate waste (particularly the waste of overproduction) by pulling work through the system based on actual demand. The “cord” in a Toyota plant which any worker can pull to stop the line when a defect is found. This is mirrored in XP’s requirement that every change passes 100% of the tests before it is integrated. By breaking down social stratification and making everyone responsible for quality and improvement (Kaizen). XP creates a more humane and productive environment than the rigid hierarchies of Taylorist organizations. [1]

Economic Foundations and the Theory of Constraints

The throughput of a software development system is governed by its narrowest bottleneck. Extreme Programming uses the Theory of Constraints to identify where work is piling up (whether in requirements, implementation, or integration) and shifts resources to address that specific constraint. [1]

The Pull Model of Development

While traditional “waterfall” processes push requirements through a series of stages, XP uses a “pull” model. Detailed specifications are pulled from stories only when implementation is about to begin; tests are pulled from those specifications; and code is pulled from those tests. This ensures that the team only produces what is needed. Minimizing the inventory of “work-in-progress” that otherwise obscures the true state of the project. [1]

Planning and the Economics of Scope

Planning in XP is treated as a negotiation over scope rather than a fixed commitment to speed or quality. The sponsor fixes the time and cost and the team provides estimates for stories. If the resulting plan is unacceptable the team and customer work together to pick the most valuable subset of stories that fits the available budget. This transparency allows the organization to manage the “time value of money” and maximize the option value of the system. [1]

Framework	Primary Focus	Structure and Roles	Predictability vs. Flexibility
Scrum	Project management process. [12]	Sprints (1-4 weeks), Scrum Master, Product Owner. [14]	High structure; predictable cycles. [15]
Kanban	Operational efficiency and flow. [10]	Visual boards, WIP limits, no fixed roles. [12]	High flexibility; continuous delivery. [11]
Extreme Programming	Technical excellence and quality. [3]	Short cycles, TDD, Pair Programming, Whole Team. [1]	Focus on technical quality; rapid adaptation. [15]

Extreme Programming in the Era of Generative AI

The emergence of Generative AI (GenAI) and Large Language Models (LLMs) in 2024 and 2025 has created a pivotal moment for Extreme Programming. [17] While some early forecasts suggested that AI might render agile methodologies obsolete. The current industry consensus is that XP’s rigor is more essential than ever to manage the “junk code” and security risks inherent in AI-generated software. [4]

The AI Pair Programming Paradigm

By 2025 tools like GitHub Copilot and Cursor have evolved from simple autocomplete systems into “AI pair programmers”. [6] This has fundamentally changed the Driver-Navigator dynamic of traditional pairing. [6]

Virtual Driver Roles: In many workflows the AI acts as the “driver,” generating implementation code while the human developer takes on the “navigator” role. Overseeing design, architecture and validation. [6]
Performance Metrics: Studies from late 2024 and early 2025 indicate significant productivity boosts in routine tasks with some teams reporting a 26% increase in completed tasks. [20] However, for experienced developers working on complex, high-quality codebases AI tools have been shown to cause a 19% slowdown due to the cognitive load of reviewing and fixing model errors. [21]
Knowledge Sharing Risks: A critical finding from 2025 research is that while AI boosts individual speed it cannot replicate the team-building benefits of human pairing. Such as shared domain knowledge and collective code ownership. 6 Teams are now advised to reserve human-human pairing for critical system boundaries and onboarding. [6]

Vibe Coding and Material Disengagement

Introduced by Andrej Karpathy in early 2025, “Vibe Coding” represents an evolution of XP’s “Simple Design” principle into a state of “material disengagement”. [22] In this paradigm, developers describe the “vibe” or intention of a feature in natural language and letting agentic models handle the mechanical implementation. [24]

Cognitive Dialogue: “Vibe coding” shifts programming from token-by-token authoring to a cognitive dialogue where intent is encoded in language and compiled by LLMs into structured systems. [23]
The PACT Framework: To maintain quality in vibe coding experts propose the PACT model: Prepare resources, Architect boundaries, Code through spells (prompts) and Test extensively. [24]
Democratizing Force: This style allows non-technical users to generate functional prototypes in a single day. Though it raises massive concerns regarding long-term maintainability and security. [23]

Agentic Software Engineering and SE 3.0

The transition from AI-assisted SE (Level 2) to Agentic Software Engineering (Level 3+) involves specialized AI teammates that contribute to planning, coding and testing as first-class collaborators. [5]

The Agentsway Methodology: New frameworks like “Agentsway” define structured lifecycles where a human orchestrator manages a suite of specialized agents (Planning Agent, Prompting Agent, Testing Agent). [26]
TDD as a Survival Strategy: In an environment where code is generated at digital speed TDD has moved from a niche practice to a mandatory survival strategy. [28] The “Lethal Trifecta”: the risk of LLMs reading sensitive data, encountering untrusted content and having external communication demands that all AI-generated code be wrapped in rigorous and human-validated automated tests. [9]

Workflow Component	Traditional XP (SE 1.0)	AI-Augmented XP (SE 2.0)	Agentic XP (SE 3.0)
Actors	Human Developers only. [1]	Humans + Copilots. 4	Human Orchestrator + AI Agents. [26]
Testing	Manual TDD cycle. [7]	AI-assisted test generation. [6]	Autonomous Agentic TDD loops. [6]
PR Process	Peer Review (Human). [3]	Human review of AI code. 4	Agentic-PRs (83.8% acceptance). [30]
Code Ownership	Collective (Human). [1]	Blurred (High human overhead). 4	Systemic (Intent-driven). [4]

Comparative Performance and Trends (2024-2026)

The integration of AI into Extreme Programming has yielded mixed results that emphasize the need for continued human judgment. [19]

Metrics (2025 Study)	Impact Outcome	Contextual Implication
Task Completion Speed	19% Slowdown (Experienced). [21]	High quality standards increase review time. [21]
Task Completion Volume	26% Increase (General). [20]	AI excels at boilerplate and routine logic. [20]
Defect Rate	25-40% Reduction (XP vs Traditional). [2]	Disciplined TDD and pairing catch bugs early. [2]
PR Acceptance Rate	91% (Human) vs 84% (Agentic). [30]	Human context is still a differentiator. [30]
Developer Favorability	72% Positive. [20]	Sentiment is tempered by real-world bugs. [20]

Scaling XP: Complexity and Consequences

Scaling XP involves managing dimensions of people, investment and risk. For large organizations the challenge is not just the number of people but the size of the organization and the consequences of failure. [1]

The Conquer-and-Divide Strategy

Faced with large programming problems, XP avoids the Taylorist approach of “divide and conquer” (where pieces are built in isolation and integrated at the end). Instead it advocates “conquer and divide”: a small team solves the core problem first. Identifies the natural fracture lines in the system and then partitions the work among autonomous teams that integrate frequently. This maintains the illusion of independent teams while ensuring systemic integrity. [1]

Safety-Critical and Security Systems

In safety-critical environments (avionics, medical) Extreme Programming is augmented with traceability and auditing. The principle of Flow suggests that auditing should not be an end-phase activity but an ongoing interaction with auditors. Teams record the lineage of every line of code back to a story and a system test, providing the documentation required for certification without halting the flow of value. In the agentic era security is integrated via “DevSecOps,” treating security policies as code and automating security checks directly into the CI/CD pipeline. [32]

Synthesis and Future Trajectories

Extreme Programming has evolved from a controversial set of Smalltalk practices into the engineering foundation of modern software development. [34] Its influence is seen in the “Shift Left” movement in DevOps, the Continuous Delivery pipelines of the 2020s and the emerging methodologies for AI-agent collaboration. [32] The core contribution of XP remains its ability to foster outstanding software development by focusing the Whole Team on common, reachable goals while celebrating human creativity. [1]

As we move toward a future defined by autonomous agents and natural language orchestration, the “Extreme” in XP will increasingly refer to the intensity of our feedback loops and the rigor of our automated verification. [5] The shift from token-driven coding to intent-driven software engineering does not mark the end of XP but rather its ultimate fulfillment: a world where developers are freed from mechanical toil to focus on architectural planning, system integration and the strategic alignment of software with human needs. [4] Reconciling humanity and productivity remains the industry’s greatest challenge and Extreme Programming remains one of its most durable solutions. [1]

Works cited

Extreme Programming Explained_ Embrace Change
What is Extreme Programming (XP) in Agile? Principles, Practices & Benefits (2025), accessed February 2, 2026, https://premieragile.com/what-is-extreme-programming-xp-agile-2025/
Extreme programming: a full guide to XP practices in 2026 - Monday.com, accessed February 2, 2026, https://monday.com/blog/rnd/extreme-programming/
Towards AI-Native Software Engineering (SE 3.0) : A Vision and A Challenge Roadmap, accessed February 2, 2026, https://www.scribd.com/document/937708646/2410-06107v1
Agentic Software Engineering: Foundational Pillars and a Research Roadmap - arXiv, accessed February 2, 2026, https://arxiv.org/html/2509.06216v2
Pair Programming & TDD in 2025: Evolving or Obsolete in an AI …, accessed February 2, 2026, https://medium.com/@pravir.raghu/pair-programming-tdd-in-2025-evolving-or-obsolete-in-an-ai-first-era-00680ce93695
pair programming evolution -from human collaboration to chatgpt llm and ms github copilot, accessed February 2, 2026, https://www.researchgate.net/publication/400092533PAIR_PROGRAMMING_EVOLUTION-FROM_HUMAN_COLLABORATION_TO_CHATGPT_LLM_AND_MS_GITHUB_COPILOT
What is Test-Driven Development (TDD)? - IBM, accessed February 2, 2026, https://www.ibm.com/think/topics/test-driven-development
2025 - Martin Fowler, accessed February 2, 2026, https://www.martinfowler.com/tags/2025.html
Software Development Methodologies in 2025 | Complete Guide - Vividtech Solutions, accessed February 2, 2026, https://www.vividtechsolutions.com/10-best-software-development-methodologies-in-2025/
Comparative Analysis of Agile Frameworks: Scrum, Kanban, Extreme Programming, accessed February 2, 2026, https://www.researchgate.net/publication/393842886_Comparative_Analysis_of_Agile_Frameworks_Scrum_Kanban_Extreme_Programming
Agile frameworks: Kanban, Scrum, and Extreme Programming (XP), accessed February 2, 2026, https://www.mytaskpanel.com/agile-frameworks-when-to-use-kanban-scrum-and-extreme-programming-xp/
Agile vs Scrum vs Kanban: Picking the Right Approach to Deliver at Scale - BairesDev, accessed February 2, 2026, https://www.bairesdev.com/blog/agile-vs-scrum-vs-kanban/
Kanban vs Scrum vs XP - TPXimpact, accessed February 2, 2026, https://www.tpximpact.com/knowledge-hub/insights/kanban-vs-scrum-vs-xp
Agile Scrum vs Kanban vs XP: Which Framework Delivers the Fastest Results? - ONES.com, accessed February 2, 2026, https://ones.com/blog/agile-scrum-vs-kanban-vs-xp-fastest-results/
Agile Vs Scrum: Key Differences Explained For Teams In 2025 - Monday.com, accessed February 2, 2026, https://monday.com/blog/rnd/agile-vs-scrum/
AI in Software Development - IBM, accessed February 2, 2026, https://www.ibm.com/think/topics/ai-in-software-development
(PDF) AI and Agile Software Development: A Research Roadmap from the XP2025 Workshop - ResearchGate, accessed February 2, 2026, https://www.researchgate.net/publication/395033842_AI_and_Agile_Software_Development_A_Research_Roadmap_from_the_XP2025_Workshop
Comments - Programming Deflation - by Kent Beck - Software Design: Tidy First?, accessed February 2, 2026, https://tidyfirst.substack.com/p/programming-deflation/comments?utm_medium=web
AI Pair Programming: How Vibe Coding Is Extreme Programming - Dino Cajic, accessed February 2, 2026, https://www.dinocajic.com/ai-pair-programming-how-vibe-coding-is-extreme-programming/
Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity - METR, accessed February 2, 2026, https://metr.org/blog/2025-07-10-early-2025-ai-experienced-os-dev-study/
Vibe coding: programming through conversation with artificial intelligence - arXiv, accessed February 2, 2026, https://arxiv.org/html/2506.23253v1
Explain in 5 Levels of Difficulty: Vibe Coding - DEV Community, accessed February 2, 2026, https://dev.to/mcsee/explain-in-5-levels-of-difficulty-vibe-coding-4f77
The PACT Framework: A Magical Guide to Principled Vibe Coding - Synaptic Labs Blog, accessed February 2, 2026, https://blog.synapticlabs.ai/pact-framework-vibe-coding-guide
Vibe Coding, Explained. AI-assisted coding, or ‘vibe coding’… | by Sunil Manghani | Electronic Life | Medium, accessed February 2, 2026, https://medium.com/electronic-life/vibe-coding-explained-e3632b8ce1bf
Agentsway — Software Development Methodology for AI Agents-based Teams - arXiv, accessed February 2, 2026, https://arxiv.org/html/2510.23664v1
(PDF) Agentsway -- Software Development Methodology for AI Agents-based Teams, accessed February 2, 2026, https://www.researchgate.net/publication/397007258Agentsway–_Software_Development_Methodology_for_AI_Agents-based_Teams
Ruby Was Ready From The Start. Notes and expanded thoughts on eXtreme… | by Obie Fernandez, accessed February 2, 2026, https://obie.medium.com/ruby-was-ready-from-the-start-4b089b17babb
Extreme Programming… Revised with Generative AI Agents | by Anton Voronko - Medium, accessed February 2, 2026, https://medium.com/@antonvoronko/extreme-programming-revised-with-generative-ai-agents-5ccacf5269e4
On the Use of Agentic Coding: An Empirical Study of Pull Requests on GitHub - arXiv, accessed February 2, 2026, https://arxiv.org/html/2509.14745v1
Developer Productivity in 2025: More AI, but Mixed Results - The New Stack, accessed February 2, 2026, https://thenewstack.io/developer-productivity-in-2025-more-ai-but-mixed-results/
The State of DevOps in 2025: Trends, Adoption, Challenges, and Future Directions, accessed February 2, 2026, https://www.baytechconsulting.com/blog/the-state-of-devops-in-2025
Best of 2025: The Future of DevOps: Key Trends, Innovations and Best Practices in 2025, accessed February 2, 2026, https://devops.com/the-future-of-devops-key-trends-innovations-and-best-practices-in-2025-2/
A Continuous Delivery reading list - Dazed & Confused, accessed February 2, 2026, https://blog.f12.no/wp/2025/01/14/a-continuous-delivery-reading-list/
Agile in 2025: Expert Predictions and Industry Trends, accessed February 2, 2026, https://www.easyagile.com/blog/agile-trends-predictions-2025
Towards Shift-Up: A Framework and a Prestudy on High-Value Activities in GenAI Native Software Development - arXiv, accessed February 2, 2026, https://arxiv.org/html/2509.24485v1

The Socio-Technical Architecture of Extreme Programming: Foundations, Philosophical Roots and the Agentic Future

The Genesis of a Paradigm: The Chrysler Creation Story

The Axiological Core: Values as the Source of Purpose

Simplicity and the Elimination of Over-Engineering