Imagine it’s a Monday morning in 2028.

A developer, instead of writing a code from scratch, is reviewing, validating, and refining code that SAP Joule has already generated—a purpose-built LLM trained on millions of lines of SAP code, delivering contextual results that support everything from intelligent code suggestions to automated unit test creation.

Integration specialist isn’t building and mapping integration flows line by line, rather they’re orchestrating an agentic workflow that maps, tests, and deploys integrations autonomously.

Process experts are no longer writing documentation for design, configuration, test scenarios or cases rather an AI-powered agent is writing this as per business context freeing up their time for key design decisions.

Now contrast that with today: standups, Jira boards, and a backlog of hand-coded programs and deliverables that take weeks to document and block valuable time in non-value-added activities.

This isn’t science fiction. It is the trajectory we are already on. I believe the teams that thrive in 2028 will be the ones that learn to embed and work with AI agents.

The forces driving this change

Several technology shifts are converging, and their combined effect on SAP delivery is groundbreaking.

Generative AI in coding is already here. According to an IDC whitepaper ‘ABAP Development in the age of AI’, SAP Joule for developers can augment developers’ efficiency by 30-40% and covers code generation, unit test generation, and code explanation—all embedded directly in the developer’s environment with no context switching required.

Agentic AI takes this further. Agentic platforms including SAP are now enabling custom agent building with capabilities including AI-assisted design, system-triggered agents, agent-to-agent (A2A) protocol support, and full Model Context Protocol (MCP) integration. These aren’t just tools that respond to prompts; they plan, decide, and act in pursuit of business goals.

Model Context Protocol (MCP) is the connective tissue that makes agentic a reality. By exposing enterprise applications through MCP servers, organisations can make their existing technology landscape truly agentic, significantly reducing process cycle times and improving operational efficiency.

Platforms are getting standardised with clean core strategy reducing complexity by emphasising configuration instead of customisation. Less custom code buried in systems means faster upgrades and patches, integrations behave predictably, and technical debt is minimised.

Role by role: What actually changes

The question isn’t whether AI will change your SAP team. The question is whether your team is changing fast enough.

Developers

Routine coding and enhancement frameworks will be largely AI-generated. Junior level coding tasks will be highly automated. What emerges is the reviewer-architect, someone who validates AI-generated code, catches performance anti-patterns, and governs what AI-assisted tools produces. New skills required are cloud native development models, RESTful Programming Model, and prompt engineering skills.

Integration Developers

New local MCP servers are reducing the grind of building connectors and mapping line by line with preferred code assistants like Cursor, Windsurf, and Claude Code while maintaining enterprise-grade governance and clean-core alignment. The new role will emerge: Agentic Workflow Architect.

Functional & process consultants

Repetitive configuration tasks like chart of accounts, pricing conditions, vendor master becomes AI-assisted at minimum, AI-automated at maximum. What grows is process intelligence: interpreting AI-generated recommendations, validating them against business intent, and governing agents running month-end close diagnostics or order-to-cash exceptions. Human contextual judgement cannot be encoded.

Testers & QA engineers

Manual regression testing will reduce dramatically. End-to-end regression testing is a prime area for agentic automation. What will be augmented is manual test execution as a primary job function. What grows is test architecture and AI governance: designing what the agent should test, curating test data, and interpreting AI-generated quality signals.

Solution architects

The architect role expands, not contracts. Architects now design not just technical landscapes but agentic landscapes: what agents exist, what they can access, how humans stay in control. MCP defines a communication protocol that separates host applications (that orchestrate tool usage) from tool servers (that expose specific capabilities), allowing for a more modular and scalable architecture. The MCP topology becomes an architectural concern. New skills: agentic system design, AI ethics, and governance frameworks.

Project managers & delivery leads

AI-assisted project management will handle routine status reporting and risk flagging. What grows is human leadership: stakeholder trust, change management, and the ability to lead teams navigating profound role transformation.

Practice leaders & centre of excellence heads

Leaders’ role becomes talent transformation at speed. They own the AI governance model. They answer to clients expecting AI-amplified delivery velocity. This role doesn’t get easier—it gets higher stakes.

Forward deployed engineers (FDE)

Traditionally SAP delivery has always been separated into thinkers (architects, SMEs) and builders (developers, integrators), which may collapse in future. One new role that will emerge is that of a forward deployed engineer, basically a very strong technical SME who has solid understanding of a business domain like finance, procurement backed by good industry knowledge. This is very much a new role that will become more prominent once AI adoption increases and the pace of change accelerates. These roles will not only shape the roadmap and create design, but also build and deploy those changes.

The new team structure: How will teams in 2028 look

The team doesn’t necessarily shrink. It reshapes. New archetypes emerge that don’t exist in most practices today:

  • Prompt Engineers – crafting high-quality AI inputs for Joule and SAP AI services
  • Agentic Workflow Designers – architecting autonomous process chains
  • MCP Specialists – configuring and governing AI-to-SAP connectivity
  • AI Governance Leads – ensuring compliance, auditability, and human oversight
  • Forward Deployed Engineers (FDEs) – design and build end-to-end solutions

As per the IDC whitepaper mentioned above, the future teams may operate at 30-40% higher productivity as AI augments capacity and allows SMEs to focus on high-value, client-focused activities. With this, the skills baseline rises significantly as every role requires AI literacy, critical judgement, domain depth, and industry knowledge.

The talent imperative: What you must do now

A challenge for a practice leader could be assuming this transition is three-five years away. Parts of it are happening right now on current projects with tool developers already having access.

In my view, 30% graduates should be trained beyond skills that AI will own in two years. Build AI-native SAP practitioners from day one: Joule, Business Technology Platform (BTP), AI Services, ABAP Cloud, and prompt engineering belong in their induction curriculum.

The 35% junior-to-mid level cohort faces the highest disruption risk and the highest retraining opportunity simultaneously. Invest in pathways toward agentic workflow design, MCP configuration, and test architecture.

The 20% senior SMEs is the most critical group. Their nuanced, hard-won industry and domain knowledge is the fuel that trains and governs AI agents. Reposition them as the intelligence layer that makes AI trustworthy, and some of these can be your FDEs.

For architects and leaders, invest now in agentic system design and AI governance. Clients will ask about this before organizations and leaders feel ready.

Closing thought

Specialised, sharper, and fundamentally different in composition—that is the SAP practice of 2028.

The human SAP practitioner is not going away. What is going away is the version of this role defined by volume, repetition, and manual execution. What replaces it is something more demanding and more interesting: practitioners who understand AI well enough to govern it, domain experts who can teach machines what good looks like, and architects who design systems where humans and agents collaborate intelligently.

The SAP practices that will lead in 2028 are already making decisions today, about hiring profiles, reskilling investments, and AI governance frameworks, that their competitors are deferring until ‘the technology matures’.

Technology is maturing. The only question is whether the workforce strategy is keeping pace.