Case Studies  /  Agile Transformation

Enterprise Agile
Transformation
— Energy Sector, Italy

85%
Sprint commitment rate
−50%
Cycle time reduction
70%
AI tool adoption
40+
People across 6 BUs

The challenge

A major Italian energy company had 6 business units delivering software in waterfall. Each BU had its own processes, tooling, and definition of "done." Cross-BU dependencies were managed through email and monthly steering committees. Cycle times averaged 12+ weeks. Stakeholder confidence in delivery timelines was low.

The brief: implement a SAFe 6 framework across all 6 BUs simultaneously, maintain operational continuity during the transition, and demonstrate measurable improvement within 8 months. The added complexity: one of the BUs had never used agile practices before, and two had attempted and abandoned previous agile transformations.

The approach

A failed agile transformation leaves scar tissue. Teams that have "done agile before and it didn't work" are harder to transform than teams that have never tried. The first step wasn't process — it was rebuilding trust.

Month 1–2
Diagnostic & coalition building
Mapped current state across all 6 BUs: processes, tooling, team structures, pain points. Identified champions in each BU. Ran individual stakeholder sessions before any group workshops. Built the case for change using their own data (cycle times, defect rates, missed deadlines).
Month 2–3
SAFe foundation & PI Planning preparation
Defined the Agile Release Train (ART) structure. Designed team topologies across BUs. Ran SAFe training for 40+ people. Set up Jira boards for all teams with shared definitions of done and consistent workflow states.
Month 3–5
First PI — structured with intensive coaching
First Program Increment with daily Scrum Master touchpoints. Identified and resolved impediments within 48 hours. Introduced flow metrics (cycle time, throughput, WIP) alongside velocity. Coached teams through their first sprint reviews with real business stakeholders.
Month 5–8
Stabilisation and AI agent deployment
Second PI with reduced coaching intensity as teams became more self-sufficient. This is when I introduced the sprint backlog pipeline: a Gemini Gem reading ceremony transcripts from Google Drive, a dedicated PBI Writer producing sprint-ready backlog items before the team sat down. Sprint capacity went up 40%, cycle time dropped from 6 days to 3. Also started the phased AI agent program for the development cycle: Copilot instruction files in Git repositories, documentation agents per microservice, a pre-development agent flagging dependency conflicts before the first commit. Ran the final ART health check and handed over to an internal Release Train Engineer.

What made the difference

Starting with their data, not a framework. Before the first workshop, I built a dashboard showing each BU's current cycle times, sprint predictability, and defect escape rate. The argument for change came from their own numbers — not from a SAFe textbook.

Designing for the resistors, not the champions. Every transformation has early adopters who make it look easy. What kills transformations is the 40% who are skeptical. I spent disproportionate time with the resistant teams — understanding their specific objections, adjusting the process where it was genuinely wrong for their context.

Metrics that connect to business outcomes. Sprint velocity is a team metric. Stakeholders care about cycle time (time from "let's build this" to "it's in production") and business value delivered per quarter. I tracked both, and reported cycle time to business leadership weekly during the first PI.

"The transformation worked because Pedro treated it as a delivery problem, not a training programme." — Stakeholder feedback, Energy Company

Results after 8 months

  • 85% sprint commitment rate — up from ~45% at the start of the transformation
  • −50% cycle time — from 12+ weeks average to 5–6 weeks across BUs
  • 70% AI tool adoption — in the BUs where AI-assisted planning was introduced
  • First successful cross-BU PI Planning — 6 BUs aligned on a shared quarterly roadmap for the first time
  • Internal RTE capacity — trained and handed over to an internal Release Train Engineer at programme close