We Need to Talk About SAFe
We have been working with the Scaled Agile Framework since its early versions. We became SAFe Program Consultants. We ran PI Planning events across the UK, Europe, the US, and Canada. We have seen SAFe deliver real value in complex, multi-team environments.
We have also seen it become a compliance exercise in organisations that adopted the framework without the mindset. And in 2026, we are seeing something new: organisations that succeeded with SAFe are now asking whether it is the right model for a world where AI is reshaping how software gets built.
What SAFe Got Right
SAFe's core contribution was solving a real problem: how do you coordinate multiple agile teams working towards a shared outcome?
- PI Planning: A structured event that brought teams together to align on shared commitments. At its best, PI Planning created genuine transparency and cross-team collaboration.
- Portfolio visibility: Lean Portfolio Management gave executives a way to see and govern the flow of work without micromanaging teams. We wrote about our experience streamlining PI Planning in FMCG manufacturing — an example of SAFe working well in a non-IT context.
- A common language: In large organisations, having shared terminology reduced confusion across hundreds of people.
- An adoption pathway: SAFe's prescriptive nature made it accessible.
Where SAFe Is Showing Its Age
SAFe was designed in a world where the primary bottleneck was coordination between human teams. AI is shifting that bottleneck.
Ceremony Overhead
SAFe's ceremony calendar is substantial. Each ceremony had a clear purpose when designed. But AI is now capable of handling much of the information-sharing that these ceremonies were built to facilitate. If an AI agent can synthesise progress across teams in real time, do you still need a two-day PI Planning event to achieve alignment? Perhaps — but the format needs to evolve.
Cadence Rigidity
SAFe's fixed Program Increment cadence (typically 8–12 weeks) assumes that planning cycles need to be synchronised across teams. In an AI-First delivery model, continuous flow and on-demand planning are increasingly viable.
Role Proliferation
SAFe introduces roles that add coordination layers: Release Train Engineers, Solution Train Engineers, Epic Owners. Each role exists to manage complexity that AI tools can now partly automate. The roles are not eliminated, but their focus shifts from coordination to strategic decision-making.
What to Keep from SAFe
- The alignment principle. The idea that teams should be aligned around shared outcomes, with visibility across the portfolio, remains essential.
- Lean Portfolio Management. Connecting strategy to execution through portfolio-level governance is one of SAFe's best contributions. This principle translates directly into OKR-driven portfolio management.
- The Inspect and Adapt cadence. Regular, structured reflection on how the system is performing is timeless.
What to Adapt
- PI Planning: Evolve from a two-day synchronisation event to a shorter, more focused alignment session supported by AI-generated context.
- System Demo: Shift from a ceremonial review to continuous stakeholder access to working software, with AI-summarised release notes.
- RTE role: Evolve from coordination manager to delivery strategist. The administrative coordination gets automated; the strategic decision-making gets elevated.
What to Move Beyond
- Prescriptive ceremony calendars. Design your ceremony cadence based on what your organisation needs, not what the framework prescribes.
- Framework certification as a proxy for competence. A SAFe certification tells you someone studied a framework. It does not tell you they can lead a transformation.
- The assumption that scaling = more structure. AI-First delivery suggests the opposite: that AI can handle coordination complexity that previously required structural solutions.
The AI-First Evolution
We position AI-First Delivery not as the enemy of SAFe, but as its evolution. The principles that made SAFe valuable — alignment, visibility, flow, continuous improvement — remain. The mechanisms change. Our Five Lenses framework (Ceremonies, Roles, Information Flow, Quality, Capacity) provides a structured way to evaluate which elements of your current scaled practice to preserve and which to redesign.
For a complete introduction to what AI-First Delivery means in practice, read our complete guide for engineering leaders.
What to Do Next
If your organisation is running SAFe today, do not rip it out overnight. Instead:
- Assess which ceremonies are adding value and which have become rituals.
- Identify where AI tools could replace manual coordination.
- Pilot AI-First practices within one Agile Release Train before scaling.
- Use OKRs to measure whether the new approach delivers better outcomes.
If your transformation has stalled entirely, our article on why agile transformations fail diagnoses the most common patterns. And if you are unsure whether SAFe is the right foundation at all, our original article Is The Scaled Agile Framework Right For Me? remains a useful starting point.
The organisations that will thrive in 2026 are not those that cling to any single framework. They are those that take the best principles from SAFe, combine them with AI-First thinking, and design a delivery practice that fits their context.
Ready to put this into practice? Book a free 30-minute consultation with McKenna Agile Consultants. No sales pitch — just a conversation about where you are and what would actually help.
