Case Study

From Weeks to Days: Reimagining Product Discovery and Engineering with SpaceXSI

Australian Leading Media Company

Challenge

A leading Australian media and publishing network, responsible for some of the country’s largest digital mastheads and entertainment destinations, sought to accelerate product delivery across its digital estate.

With major strategic initiatives underway, including the unification of its digital games and interactive experiences, the organisation partnered with Fabric Group to deliver a focused four-week Proof of Value. In collaboration with SpaceXSI, Fabric embedded an orchestrated, AI-native software delivery lifecycle across product, design and engineering teams, accelerating delivery across key strategic initiatives.

Despite strong engineering capability, the organisation faced four critical bottlenecks across its delivery pipeline:

  • Extended Discovery Cycles: Synthesising customer research, analysing competitors, and translating product strategy into engineer-ready requirements took weeks of manual effort.
  • Delivery Friction in Complex Environments: Heavy architecture tasks, such as multi-repository restructures and Kafka event-stream processing, tied up senior engineering capacity on repetitive boilerplate.
  • The Review Bottleneck: Accelerating code production without established guardrails risked overwhelming senior reviewers with sprawling pull requests lacking architectural consistency.
  • Design-to-Code Handoff: Teams faced persistent friction between Figma and code, where manual interpretation created divergence between design sprint concepts and front-end execution.
No items found.
PRocess

Fabric embedded directly within product, design, and engineering streams to implement repeatable workflows and delivery guardrails structured around three core pillars:

1. Plan & Design: Discovery to Endorsed Scope

Applied AI tooling to upstream product definition to compress cycle times from weeks to days:

  • Accelerated Competitor UX Intelligence: Cut competitor analysis from over a week of manual research down to 2 days using design-reference integrations.
  • Design Sprint to PRD: Turned prioritisation boards and wireframes into an endorsed Product Requirement Document (PRD v1.1) and a playable prototype, advancing delivery by months.
  • Design System Skills: Created custom commands that read product repos and the design system, binding components to design tokens with “no raw values” (documenting 18 components with one prompt each).
  • High-Fidelity UI Extraction: Converted Figma frames into clean production-ready code with minimal manual interpretation.

2. Build: Orchestrated Multi-Repo Engineering

Transitioned teams from simple syntax autocomplete to autonomous background workflows and multi-repo orchestration:

  • Autonomous Cloud Agents: 72 team members ran 1,335 background agent sessions, handling routine scaffolding and migrations asynchronously.
  • Legacy Modernisation at Scale: Prepared a major platform framework upgrade (previously estimated at 6 months of full-team work) for review in approximately 72 hours across 190 changes and 380 automated review runs.
  • Rapid Feature Scaffolding: Reduced idea-to-first-PR time from 21+ days to 3 days on complex interactive rebuilds, achieving bug-bash readiness in 4 weeks.

3. Review & Test: Codified Senior Standards

Embedded automated pre-checks and architectural guardrails so code quality scaled alongside velocity: 

  • Codified Architectural Conventions: Encoded senior engineering patterns and Storybook guidelines into reusable agent skills to enforce consistency before human review
  • Multi-Model Consensus: Deployed workflows where frontier models independently reviewed PRs to combine findings into a unified verdict, while keeping senior engineers at the approval gate.
  • Smaller, Safer PRs: Adopted bite-sized PR templates and fine-tuned review bots, eliminating merge conflicts and keeping human reviews focused on architecture.
No items found.
Results

The four-week engagement proved substantial time savings on live deliverables alongside rapid enterprise-wide adoption.

Measurable Time Saved on Real Work

  • Product Specification Tasks: Cut from 1 day to ~30 minutes (90%+ cycle time reduction).
  • Multi-Repo Restructuring: Completed in 3 hours (down from an estimated 3 days).
  • Competitor UX Research: Reduced from ~1 week to 2 days.
  • Code Integrity: Zero PR errors or merge conflicts across tracked initiatives.

Rapid Enterprise Adoption

  • 103 of 140 team members engaged actively across product, design, and engineering (with 40 of 42 core stream members active in Week 4).
  • 56 contributors made their very first AI-written commit.
  • Usage Expansion: Weekly agent requests grew from 3.2k to 7.4k, while autonomous background workflow users expanded from 2 to 19.

Team Sentiment

  • 88% of engineers reported being measurably more productive.
  • 73% of engineers preferred the AI-native environment over their previous development setup.
  • High Overall Satisfaction: 77% of surveyed participants were satisfied or very satisfied (with 0% reporting dissatisfaction).
No items found.
No items found.

Related Content

No items found.
No items found.