Culture & Methods Trends 2026: The Human Side of AI Engineering

AI Adoption and Professional Perspectives

• Decisions regarding AI adoption, architectural trade-offs, and team collaboration methods are currently shaping the long-term trajectory of software systems 0s.
• QCon San Francisco is scheduled for November 16-20, featuring over 60 speakers across 12 tracks focused on practical, production-level insights from senior practitioners 15s.
• Vanessa Fikola serves as an engineering leader and sociotechnical architect, focusing on addressing systemic problems across both technical and leadership domains 42s.
• Ben Linders operates as a one-person company specializing in the social aspects of sociotechnical team collaboration, including people safety, reflection, and learning 52s.
• Philip Mortimer is a machine learning engineer with a background in finance and fintech, currently specializing in document intelligence 1m1s.
• Yinka is a software engineer with a background in banking, currently working as an individual contributor within the HR tech industry 1m11s.
• Raph, a senior director in the tech industry with experience dating back to the 1980s, focuses on cross-border and cross-time-zone team collaboration, feedback loops, and transitioning to an AI-first approach 1m18s.
• Craig Smith has 20 to 25 years of experience in the agile and culture and methods community and currently works as a delivery leader within the government sector 1m33s.
• Shauna Martell is a principal engineer in the fintech space who focuses on fostering staff-plus engineers within organizations 1m42s.
• Shane Hasty leads the culture and methods division at InfoQ, focusing on leadership development and the human side of sociotechnical systems 1m50s.

Strategic AI Integration and Risk Assessment

• A primary consideration for organizations today is determining the appropriate level of maturity and the specific extent to which AI should be integrated into their systems 2m15s.
• There is a need to move away from an "everything or nothing" approach to AI adoption and instead evaluate the specific risks and appropriateness of AI for different software, companies, and groups of people 2m25s.
• Applying traditional risk and impact assessment frameworks to software remains essential, as different pieces of software require varying levels of human oversight depending on their specific maturity and risk profiles 0s.
• The current trend of applying AI to everything should be tempered by an evaluation of whether AI serves as a genuine catalyst for improvement in each specific context 0s.

Psychological Safety and Cognitive Load in Engineering

• Collaboration patterns between teams and organizations are shifting, necessitating a focus on psychological safety and transparency to ensure individuals feel secure enough to raise issues or identify opportunities 25s.
• The successful adoption of AI is linked to the capability of team members to work with these tools while maintaining the psychological safety required for effective communication 25s.
• While psychological safety has been a focus since 2014, recent global developments and the impact of the COVID-19 pandemic have created setbacks, making it an ongoing and critical challenge 55s.
• Hyper-velocity and compound engineering, which involve managing multiple work streams and AI agents simultaneously, significantly increase the cognitive load on human engineers 1m20s.

Managing Code Volume and Development Velocity

• Traditional development processes, such as the standard pull request model, are becoming ineffective because humans cannot realistically review the sheer volume of code generated by AI 1m20s.
• Research suggests that a 300% increase in code production is often accompanied by a 400% increase in bugs 1m45s.
• GitHub pull requests are projected to grow from 1 billion per year to 14 billion by 2026, representing a 14-fold increase that necessitates new systems for managing and understanding software development 1m45s.
• Software development is currently undergoing a redefinition as teams move faster and produce larger pull requests, necessitating a continued focus on the value being delivered to users 0s.
• Despite the ability to generate code rapidly, organizations must evaluate whether their output provides the expected return on investment, particularly given the high costs associated with token usage 0s.
• There is a need for caution regarding the heavy reliance on upstream model providers that currently subsidize capabilities, as organizations must consider their long-term strategies for these dependencies 0s.

Historical Practices and Modern Maturity Gaps

• Historical software engineering methods, such as static analysis and guardrails, remain relevant and provide a necessary foundation for enabling modern teams 42s.
• Many established practices have been rebranded, and studying these historical approaches is essential for maintaining high code quality and confidence in software delivery 42s.
• Jim Highsmith has noted that organizations that failed to implement agile principles will likely experience catastrophic failure when attempting to adopt AI 42s.
• Many organizations have not yet embedded fundamental agile principles, leading to challenges in managing larger pull requests and maintaining confidence in software releases 1m24s.
• The rapid pace of AI adoption has created a maturity gap, as many organizations attempt to integrate agentic workflows without having established the necessary fast feedback loops or observability 1m24s.
• A significant divide exists between organizations that have successfully embraced modern practices and those that continue to operate using outdated software development models from 1995 1m55s.
• New challenges have emerged regarding how to effectively embed humans in the loop, manage collaboration with AI agents, and ensure quality when the same machine is responsible for building, testing, and deploying code 1m55s.
• The current evolution of software development is occurring at a pace that exceeds the development of supporting practices for both people and technology 1m55s.

Defining Value and Quality in AI-Generated Software

• While writing code has become significantly easier, the primary challenge remains ensuring that the software actually solves the intended problems for customers 2m35s.
• Organizations frequently struggle to measure whether their experiments and shipped products actually solve intended problems, a challenge that may be exacerbated by the scale of AI implementation 0s.
• The ease of generating code and the commoditization of knowledge raise questions regarding how to define and verify the delivery of genuine value to customers 15s.
• There is a concern regarding the production of "code slop," analogous to the Harvard Business Review's concept of "work slop," which refers to AI-generated content that complicates processes for those further down the value stream 30s.
• The role of the engineer is shifting from primarily writing and reviewing code to acting as a custodian, particularly as AI agents generate massive chunks of code 42s.
• Many engineers are transitioning into platform engineering, focusing on developing systems that allow agents to write code while enabling humans to review it and facilitate contributions from colleagues 42s.

Trust, Guardrails, and Evolving Team Structures

• While AI adoption is increasing, there is a notable contradiction in that trust in AI outputs remains low 1m8s.
• Engineers are expected to evolve their focus toward increasing trust in AI by building systems, guardrails, and frameworks that ensure proper business outcomes 1m8s.
• Traditional team structures, such as cross-functional "two-pizza teams," are evolving into smaller "one-pizza teams" consisting of fewer humans supported by AI tools or swarms of agents 1m25s.
• Future team compositions will likely depend on the risk level of the project, with business-critical tasks potentially requiring human teams superpowered by agents, while lower-risk projects might be managed by a single product manager and one developer 1m45s.
• The evolution of software development requires ongoing consideration of how many human interactions are necessary and which specific human skills are required to maintain and evolve software systems 1m45s.
• Building software for humans requires a different approach than writing software for machines, necessitating a clear understanding of goals such as quality, return on investment, and the management of cognitive overload before deciding on team size or hyper-automation strategies 0s.
• Failing to be honest about how to manage new technology can lead to risks, such as the hyper-generation of systems that no one understands, which can cause organizational collaboration models to backfire 0s.

Blurring Roles and the Future of Collaboration

• The traditional boundaries between roles like product managers, designers, and engineers are blurring, leading to a shift where these individuals are increasingly categorized simply as "builders" 42s.
• In modern organizations, product and design team members are increasingly committing code, while engineers are shifting toward providing the underlying primitives, architectural designs, and APIs that grant others access to data 42s.
• The focus of organizational structure is shifting away from the traditional concept of "teams" toward finding more effective ways to facilitate collaboration, whether through one-on-one interactions, human-AI agent partnerships, or broader organizational models 1m25s.
• The "engineer-manager pendulum" is causing managers to return to writing code and delivering features, further blurring the lines between roles within engineering and across different verticals 1m50s.
• There is a growing trend toward smaller teams or even one-person teams, prompting questions about how technology can be used to accelerate and improve collaboration rather than just increasing individual output 1m50s.
• A potential danger of one-person teams is the loss of the human aspect of work, as the social interaction and enjoyment derived from working with colleagues are significant components of the professional experience that AI agents cannot replicate 2m25s.

Shifting Trends in AI Usage and Team Dynamics

• A June 2026 Harvard Business Review article indicates that the primary use of AI among the general public is for therapy and companionship, while astrology and tarot readings have emerged as the ninth most common use 0s.
• Technical applications of AI, such as autonomous agent generation and code improvement, have seen a shift in rankings; while they were the fifth and eighth most common uses in 2025, they moved to 78th and 69th respectively by 2026, reflecting a broader adoption of AI tools beyond the technology industry 22s.
• Traditional software development roles, such as product owners, scrum masters, and technical leads, are beginning to disappear as individuals increasingly take on multiple functions within "one-person teams" 55s.
• Organizations are shifting toward larger teams of 25 to 30 people, which function as collections of independent one-person units rather than collaborative groups 1m20s.
• The collaborative nature of software development is being undermined as team members use AI to generate and summarize large volumes of documentation, leading to a cycle where AI is used to manage the output of other AI tools 1m35s.
• There is a concern that the reliance on AI tools is causing teams to lose the ability to engage in meaningful collaboration, challenge one another, and integrate diverse viewpoints and expertise 2m6s.
• The ease of using AI to generate functional code can create a false sense of capability, as it does not necessarily provide the professional judgment required to determine if a system will be scalable or maintainable in the long term 2m45s.
• An example of the risks associated with over-reliance on AI involves a tax company that attempted to rebuild a system using less experienced staff and AI, which ultimately resulted in project failure 2m35s.

Human Judgment and AI Limitations

• AI models currently demonstrate high proficiency in refactoring code, with expectations for further improvement over the next two years 0s.
• There is a concern regarding whether the high volume of output generated by AI is sustainable and whether essential human skills are being neglected in the process 0s.
• Senior leaders are increasingly using AI tools to build solutions, often implementing guardrails and harnesses to manage the output 0s.
• A distinction is drawn between the "plasticity" of a junior engineer, who learns from mistakes and develops over time, and the static nature of an AI model that requires explicit instructions to avoid repeating errors 0s.
• AI models may lack the ability to holistically consider architectural consistency or the appropriateness of library choices in the same way a human engineer would 0s.
• There is a risk that the industry is returning to a "rock star" developer culture, which undermines the previously established importance of team diversity and varied perspectives 0s.
• Because AI agents are trained on the same corpus of knowledge, the reliance on these tools may lead to a loss of personal and intellectual diversity within engineering teams 0s.

Career Paths and Junior Engineer Development

• The evolution of AI necessitates a reevaluation of career paths and the methods used to train junior engineers to become senior engineers 1m25s.
• There is uncertainty regarding the future level of interest in traditional junior engineering roles, as the criteria for these positions may be shifting 1m25s.
• Junior engineers may increasingly serve as gatekeepers who validate work from UX and product teams to ensure it meets standards for scalability and codebase integration 1m25s.
• The role of a gatekeeper requires advanced judgment and an understanding of product direction, which traditionally represents a later stage in an engineer's professional development journey 1m25s.
• Achieving a certain level of professional maturity is considered a prerequisite for effectively utilizing entry-level AI guardrails 0s.
• Currently, AI acts as a significant force multiplier for senior engineers who possess established judgment and experience, whereas for junior engineers, the impact of AI as a force multiplier may not always be beneficial 12s.

Skill Evolution and Professional Growth

• Within six months to a year, as AI models become more capable, the primary professional skill for engineers will shift toward the ability to effectively utilize AI and manage multiple agents 32s.
• Younger generations, characterized by greater mental plasticity, are expected to adapt to these new AI-driven skills more quickly, potentially surpassing more experienced engineers in the near future 52s.
• A significant challenge remains in developing the ability to ask the right questions regarding the product being built and the specific problems being solved, a process that traditionally involves a long journey of micro-realizations 1m12s.
• It is possible that future AI agents could be designed to coach individuals in making sound product decisions, though the personal journey of professional growth remains a point of interest 1m35s.
• While the short-term transition may be messy, the long-term method of learning for engineers will change as the necessity for manual implementation skills, such as writing specific programming languages, decreases 1m48s.
• The reduction in time spent on manual coding tasks will free up mental bandwidth, allowing engineers to focus on higher-level tasks like structuring questions and identifying technical debt 2m10s.
• Career timelines are expected to compress, with individuals reaching leadership roles faster as they manage agents to deliver complex features 2m25s.
• Despite changes in the mechanics of coding, the fundamental requirements for senior leadership—such as maintaining product vision, understanding when to pivot, and exercising sound judgment—are expected to remain largely unchanged 2m55s.
• Leadership expectations remain consistent over time, as the core requirement for leaders is the ability to effectively utilize available tools and foster those same capabilities in others 0s.

Transformation of Professional Roles and Responsibilities

• While the fundamental values expected of leaders are not changing, the structure of professional roles is expected to undergo significant transformation 35s.
• Traditional role distinctions, such as those between architects, developers, tech leads, and managers, are becoming obsolete 55s.
• Modern engineering roles require a broad range of skills rather than deep specialization in a single area, moving beyond the traditional "T-shaped" professional model 1m8s.
• Proficiency is now measured by the breadth of an individual's skills, and the ability to combine technical expertise with people and leadership skills is essential for remaining competitive in the job market 1m35s.
• A significant challenge in this new landscape is managing cognitive overload while maintaining a broad skill set 2m0s.
• The new criteria for professional responsibility are shifting from a focus on individual output to a focus on the scope of what an engineer can be accountable for while utilizing available tools 2m15s.
• Organizations faced difficulties with career path definitions prior to the emergence of AI, but AI has exacerbated these existing structural problems 2m35s.

Creativity and Critical Thinking in the Age of AI

• There is a growing concern regarding the decline of critical thinking at a societal level, which raises questions about the impact on the software industry 2m55s.
• Creativity is expected to become increasingly important in the age of AI, as AI is capable of producing syntactically correct and high-quality code but may lack the ability to generate creative solutions to complex problems 3m10s.
• Historical technological revolutions, such as the invention of photography, have historically enabled creation at unprecedented speed and scale, leading to significant creative movements like impressionism and abstract art 0s.
• Parametric design in architecture allowed professionals to specify constraints—such as load-bearing requirements and shape—rather than manually determining structural details, which resulted in an explosion of architectural creativity 0s.
• The integration of AI is expected to mirror these historical trends, fostering a creative explosion by enabling a broader range of people to build applications and websites 0s.
• Constraints can enhance creativity by reducing the burden of excessive choices, allowing individuals to focus on solving actual problems rather than debating technical details like language or library selection 0s.
• As AI models improve and users gain more experience with these tools, a new wave of problem-solving and creativity is expected to emerge 0s.

Foundational Expertise and Human Oversight

• The ability to generate code or build applications does not equate to the expertise required for specialized, high-stakes tasks, such as developing secure transaction processors for banking 0s.
• Critical thinking has long been a foundational component of professional skill sets across various fields, including testing and project management 0s.
• Despite the accessibility of AI agents, there remains a critical need for specialized expertise in areas like secure architecture and cloud infrastructure, such as AWS 0s.
• AI tools should be viewed as a means to enhance collaboration and creative thinking, rather than a replacement for the core skill sets and deep domain knowledge held by experienced professionals 0s.
• Maintaining core professional skills such as business analysis, project management, software engineering, and architecture remains essential to ensure AI tools are directed appropriately 0s.
• Humans must remain in the loop to evaluate AI outputs, provide necessary constraints, and craft precise prompts to achieve desired results 0s.
• Domain knowledge and a deep understanding of customer needs are critical for effectively collaborating with both people and AI systems 0s.
• Direct communication with customers, such as gathering feedback on new features and observing product usage, remains the most effective way to understand business requirements 0s.
• Professionals must cultivate the ability to accept contradictory information, acknowledge when they are wrong, and adjust their behavior accordingly 0s.

Ethics, Accountability, and Environmental Impact

• The concept of "just because we can doesn't mean we should" serves as a foundational principle for exploring the ethics of products and engineering practices 0s.
• Ethical engineering involves ensuring the protection of users, colleagues, and society, and building systems that do not result in negative consequences 0s.
• A significant concern in the current usage of AI is the potential loss of accountability, as increased production capabilities can lead to a decreased understanding of the systems being built 0s.
• Professionals must maintain a sense of responsibility for the code they deliver, regardless of whether it was generated by AI 0s.
• To ensure true accountability, organizations must implement clear, well-structured controls and testing harnesses around AI-generated systems 0s.
• Ethical considerations extend to the broader impact of computing, including the consumption of resources such as electricity and water 42s.
• There is significant uncertainty regarding how to utilize AI tools without posing long-term risks to civilization, given the high cost and potential impact of these technologies 0s.
• While AI presents risks of massive damage, it also offers potential upside, such as solving complex problems like nuclear fusion or advancing healthcare through the treatment of cancer 35s.
• There is a concern that individuals and organizations often prioritize scaling and the use of powerful tools without sufficient intentionality or consideration of the environmental consequences 53s.
• Historical efforts, such as those involving UT, emphasized the importance of being aware of the environmental footprint of infrastructure and services 1m15s.
• Token consumption currently serves as a proxy for measuring the impact of work, though awareness of this impact is often limited to instances where quotas are reached or resources are exhausted 1m30s.
• The startup Mural Watt utilizes a pricing model based on power consumption rather than token counts, which serves to expose the environmental impact of AI usage to the user 1m50s.
• The current phase of AI development coincides with a critical period for global carbon reduction targets, raising concerns that the rapid production of new features is leading to increased energy consumption and carbon emissions 2m25s.

Workplace Anxiety and Human-Centric Engineering

• Research, such as a study conducted earlier in the year, indicates that AI tends to intensify individual workloads rather than reducing them 2m45s.
• A significant portion of the workforce, estimated at 70% to 80%, expresses anxiety regarding the potential for their roles to be affected or replaced by AI 3m0s.
• The combination of increased anxiety, higher workloads, and rapid technological deployment prompts questions about whether current AI implementation strategies align with desired human outcomes 3m15s.
• While there is hope for developing better long-term methods for collaboration and system usage, the current short-term impact on individuals and teams is characterized by intense pressure 0s.
• The pace of engineering is accelerating, leading to a phenomenon of "hyperengineering" where individuals feel compelled to manage multiple abstract concepts simultaneously 15s.
• Context switching remains a high-cost activity that humans struggle with, raising concerns about the long-term impact on individual capabilities as people attempt to maximize efficiency 45s.
• AI output quality varies based on the user's ability to evaluate results, with some individuals relying too heavily on AI outputs without sufficient critical oversight 1m12s.
• There is a potential risk that the drive for professional growth and learning may diminish as reliance on AI tools increases 1m30s.
• A desire for greater mindfulness is expressed regarding the use of AI, emphasizing the need to balance business growth with respect for people and the environment 1m45s.
• It is important to maintain a focus on the human aspect of work, including checking on colleagues to ensure they feel comfortable and supported during periods of rapid change 2m10s.
• Despite the ability to build software quickly, there is a need for a greater focus on ensuring that the products being built are robust and valuable 2m35s.

Maintaining Human Values in Technological Transformation

• The current technological shift is compared to the Industrial Revolution, noting that while some traditional skills are becoming less valuable, the core mission of building products that meet real needs remains constant 2m48s.
• Compassion, humanism, and intentionality are identified as essential values to maintain while navigating the transformation of the industry 3m5s.
• Senior leaders have a responsibility to mentor those entering the field and to ensure that products are developed with minimal negative impact on society 3m20s.
• Artificial intelligence is recognized as an incredible tool, yet there is a need to maintain focus on the fundamental aspects of professional roles that should remain unchanged 0s.
• There is a desire for the professional conversation to shift away from an exclusive focus on outputs and metrics toward prioritizing collaboration, creativity, and the human elements of work 7s.
• Emphasizing the human element is intended to help professionals work together more effectively and increase their enjoyment of their work 7s.
• The progress that has led to current technological advancements is attributed to human collaboration and the "human in the loop" concept 23s.
• Innovation is defined as being broader than just the capabilities of AI, requiring humans to innovate work practices to improve the results derived from AI tools 23s.
• True innovation is driven by people collaborating and communicating with one another rather than relying solely on the outputs generated by AI tools 23s.
• Within the engineering culture space, the core focus remains on the people being served by the work being performed 45s.