How AI Programming Software Changes Project Management

Why Traditional Backlogs are Obsolete and Project Managers Must Redefine Their Role

English edition — originally published in German as Wie sich das Projektmanagement durch KI-Programmiersoftware verändert.

The Speed Revolution: When Development Velocity Explodes

AI-powered programming tools are not just accelerating software development; they are fundamentally changing traditional project management. The speed at which AI generates code forces a radical rethinking of backlogs, sprint planning, and the role of the project manager. What we are currently experiencing is not an evolution, but a revolution in the conception, planning, and execution of software projects.

At Deep Impact AG, we have been leaders in AI integration since 2017, working with clients from Julius Bär to our own ventures like AVA-X and Deepico. Through this experience, we have observed a paradigm shift that many companies are only now beginning to grasp: AI not only accelerates development but also fundamentally alters the meaning of project management. This article explores these changes and critically questions whether the traditional role of the project manager can survive this transformation.

The most immediate and dramatic impact of AI programming tools is speed. Not just 10–20% faster, but in many cases, 50–70% faster. GitHub reports that developers using Copilot complete tasks 55% faster, but our experience at Deep Impact suggests this still underestimates the actual speed gains for certain types of development work.

This acceleration has a cascading effect: when individual coding tasks are completed faster, sprint capacity dramatically increases. What used to be worth a two-week sprint can now be done in a few days. The traditional rhythm of software development – plan for two weeks, develop for two weeks, review, repeat – suddenly seems antiquated and inefficient.

The consequence? Development packages become significantly larger. Teams can tackle more ambitious features in a single iteration. The scope that would previously have been spread across three or four sprints can now be delivered in a single sprint.

The End of the Traditional Backlog

Here, the transformation becomes radical: the massive, meticulously maintained backlogs that previously formed the cornerstone of agile project management are becoming obsolete.

Why? Because AI implementation translates requirements into functional code with remarkable accuracy. When a project manager or product owner describes what needs to be developed, AI tools can generate implementations that are 70–80% correct on the first try. The remaining 20–30% are refinements, edge cases, and business-specific logic – but the foundation is immediately in place.

This fundamentally changes the value proposition of detailed backlog planning:

• Weeks of breaking down features into detailed user stories become superfluous when AI can implement entire sets of functions within a few days.
• Extensive estimation ceremonies (Planning Poker, Story Pointing) lose relevance when development time is drastically reduced.
• Maintaining a six-month roadmap with detailed tickets makes little sense when requirements can be transformed into working software within a few hours.
• The backlog refinement process, which traditionally consumes 5–10% of team capacity, represents an effort that no longer delivers proportional value.

The new reality: companies need smaller, more strategic backlogs that focus on business outcomes rather than implementation details. The "just-in-time" backlog, where requirements are defined days or even hours before implementation, becomes not only practicable but optimal.

The Uncomfortable Truth: Project Management Will Shrink Drastically

If development is faster and backlogs are smaller, what happens to project management? The answer, which many in this profession are reluctant to face: the traditional project management workload shrinks drastically.

Consider what happens to classic PM activities:

• Sprint Planning: Shortens from half-day ceremonies to brief alignment meetings
• Progress Tracking: Occurs almost in real-time, rather than requiring daily stand-up meetings and burn-down charts
• Risk Management: Shifts from schedule risks (which decrease with speed) to quality and architectural risks
• Resource Allocation: Simplifies when teams can achieve more with the same number of staff
• Stakeholder Reporting: Requires fewer forecasts and predictions when delivery cycles are measured in days

At Deep Impact, our Design Sprint methodology, which already delivers functional software in 4 days, has evolved further with the integration of AI. The effort for project management has been reduced by approximately 40%. Tasks that previously required the dedicated attention of the project manager now occur organically, as the implementation cycle has been so significantly shortened.

> The provocative question: If AI reduces the need for traditional project management activities by 40 to 60%, do companies need fewer project managers? Or do project managers need to take on fundamentally different tasks?

The Redefined Role: From Task Coordinator to Business Process Architect

The answer is "the latter" – but the required transformation is significant. The project managers who succeed in an AI-powered development environment will be those who shift from coordinating tasks to shaping business outcomes.

From Technical Coordination to Business Process Expertise

If AI can generate code from requirements with 70–80% accuracy, the bottleneck shifts from implementation to requirements definition. However, these are not requirements in the traditional software sense, but requirements that reflect a deep understanding of business processes, user flows, and organizational dynamics.

The new project manager must be able to:

• Map complex business processes and identify where software can create real value
• Translate business objectives into system behavior that AI can implement
• Understand industry-specific workflows well enough to precisely define requirements
• Recognize which business problems are suitable for AI-accelerated solutions and which require human judgment
• Bridge the gap between executive strategy and technical implementation without getting lost in implementation details

This is a fundamental shift from coordinator to consultant – from someone who manages schedules and tracks progress to someone who understands business operations as well as a business analyst or process consultant.

The Rise of Conceptual Work

Since implementation speed is no longer a constraint, conceptual work becomes the critical path. The project manager of the future will spend their time on:

• System architecture and integration design – how different components fit together
• Defining business logic – the rules, exceptions, and edge cases that define how the business operates
• User experience design – not visual design, but the flow and logic of user interactions
• Data modeling – what information needs to be captured, stored, and analyzed
• Defining quality criteria – what "done" means from a business outcome perspective, not just a technical one

Notice what's missing from this list: tracking velocity, managing burn-down charts, facilitating stand-up meetings, updating Jira boards. While these activities still occur, they are no longer the area where project managers create the most value.

The Crucial Challenge: Is This Change Realistic?

This vision of the transformed role of the project manager is compelling – but is it achievable? Several significant challenges deserve close consideration:

Challenge 1: The Skills Gap

Most project managers today have been trained in coordination and process management, not business process analysis or conceptual system design. The shift from "keeping projects on track" to "designing business solutions" requires completely different skills. Can existing project managers be retrained for this role? Or do companies need to hire a new generation of hybrid business-technical leaders?

The reality is that many current project managers will struggle with this transition. Business process expertise requires industry knowledge, analytical rigor, and systems thinking that cannot be learned in a few training courses.

Challenge 2: The Assumption of Accuracy

The argument that AI generates "70–80% correct code" assumes clear, unambiguous requirements. However, many software projects fail not due to implementation errors, but because requirements were misunderstood or business needs were poorly articulated. AI does not solve this problem – it can even exacerbate it by making it easier to develop the wrong thing quickly.

If the quality of requirements remains poor, companies will simply accumulate technical debt faster. Project managers who do not deeply understand business processes will greenlight AI-generated implementations that miss their mark – creating more work rather than less.

Challenge 3: The Complexity of Integration

Modern software projects rarely involve the development of isolated applications. They require integration with existing systems, APIs, databases, and third-party services. AI is excellent at generating standalone code but struggles with complex integration scenarios that require deep system knowledge.

This means that while individual features can be implemented faster, the integration effort – and the project management required for it – remains significant. The vision of a drastically reduced PM workload may apply to greenfield projects, but less so for enterprise environments with complex legacy systems.

Challenge 4: The Human Factor

Software projects are inherently collaborative, involving developers, designers, stakeholders, and end-users. Even if AI accelerates coding, the aspects of human collaboration – alignment, communication, conflict resolution, expectation management – remain time-consuming. Project managers must continue to facilitate these interactions, and AI does not reduce this workload.

The interpersonal and political dimensions of project management can become even more important as technical implementation accelerates. Stakeholders accustomed to slow delivery times might struggle to keep pace with AI-accelerated schedules, creating new coordination challenges.

Deep Impact's Perspective: Pragmatic Optimism

At Deep Impact, we believe that the transformation of project management is real and inevitable – but not immediate and not everywhere. Our experience integrating AI since 2017 in projects ranging from AVA-X to enterprise applications for financial institutions has provided us with valuable insights.

> Our philosophy: AI is not a product – AI is a solution. The effectiveness of AI-accelerated development depends entirely on how well you understand the problem to be solved.

We have seen projects where AI reduced development time by 60%, and projects where the gains were minimal. The difference? The quality of understanding of business processes and the definition of requirements. When project leaders deeply understand the business context, AI is transformative. If they do not, AI simply delivers the wrong solution faster.

What Does This Mean for Companies?

• Invest in business process training for technical leaders – this is now as important as technical skills.
• Recognize that not all project managers can make this transition; some roles will disappear.
• Create hybrid roles that combine business analysis, system architecture, and project coordination.
• Reduce backlog effort, but maintain strategic roadmaps focused on business outcomes.
• Accept that AI acceleration brings new coordination challenges, even as it reduces others.

Practical Steps for Project Managers

If you are a project manager facing this transformation, here are concrete steps to prepare:

1. Deepen your business domain expertise – Become an expert in the business domain you operate in. Understand workflows, regulations, user needs, and competitive dynamics at a level comparable to experienced business stakeholders.

1. Learn conceptual modeling – Develop skills in business process modeling, data modeling, and system architecture. These are now core competencies of a project manager, not optional extras.

1. Reduce administrative backlog effort – Experiment with smaller, more strategic backlogs. Measure whether reduced planning effort actually accelerates delivery or causes chaos.

1. Test AI tools yourself – Understand firsthand what AI can and cannot do. Use tools like GitHub Copilot or ChatGPT to see where they excel and where they fall short.

1. Shift metrics from velocity to outcomes – Stop tracking story points and start tracking business outcomes. Did the software solve the problem? Did users adopt it? Did it create value?

1. Collaborate with business analysts – If your company has business analysts, learn from them. If not, acquire their skills yourself.

1. Accept the evolution of your role or exit – Be honest about whether you can make this transition. Some project managers will succeed in the new paradigm; others should transition to roles that better suit their strengths.

Conclusion: Evolve or Become Obsolete

The project management profession is at a turning point. AI-powered development is not a future scenario – it is changing projects today. Traditional backlog management, sprint ceremonies, and progress tracking are losing relevance as development cycles become shorter and implementation accelerates.

The project managers who survive – and succeed – will be those who reinvent themselves as business process architects: professionals who understand business operations so well that they can define requirements that AI can implement precisely. This is a more demanding role than traditional project management and requires skills that many current PMs do not possess.

The transition will be difficult. Companies will need fewer project managers, but the remaining ones must be significantly more competent. This is not a pleasant message, but an honest one.

At Deep Impact, we are committed to navigating this transformation together with our clients – honestly, pragmatically, and successfully. The future of software development is AI-powered, but the future of project management is human expertise applied in fundamentally new ways.