The Problem
What It Looks Like
- • Codebase is 5-15+ years old, written in outdated frameworks
- • Simple changes take weeks instead of days
- • Fear of touching certain parts ("if it works, don't touch it")
- • New developers take 3-6 months to become productive
- • Integration with modern tools and APIs is painful
What It Costs
- 40-60% of engineering time spent on maintenance
- 3-5x longer development cycles
- Higher defect rates from lack of test coverage
- Difficulty hiring developers
- Security vulnerabilities in unmaintained dependencies
The Real Risk: Legacy systems don't just slow you down — they create an invisible ceiling on what's possible. You can't adopt AI-native development, can't integrate modern services, can't respond to market changes quickly.
The Solution Approach: AI-Assisted Modernization
Don't throw away 15 years of code. Use AI to understand what the system does, then rebuild it intelligently. This is reverse context engineering — making the implicit explicit.
Reverse Context Engineering (2-4 weeks)
Understand what the legacy system actually does before you rebuild it.
- • Code Archaeology Agent: Maps 100k+ lines of code to business concepts
- • Dependency Mapper: Visualizes hidden system architecture
- • Test Extraction: Reverse-engineers specs from existing tests
- • System map: Data flows, components, boundaries
- • Implicit specs: What the code actually does
- • Risk matrix: High-risk, low-complexity areas
Why this works: AI can read code 10x faster than humans. It finds patterns, clusters related functions, and explains intent. You get clarity without manual archaeology.
Target Architecture & Foundation (4-8 weeks)
Design the new system and build the scaffolding. AI agents accelerate decisions and prevent wrong choices.
- • Proposes target architecture based on current constraints
- • Evaluates microservices vs. modular monolith vs. strangler fig
- • Documents decision rationale for team understanding
- • Code Generation Agent scaffolds new application structure
- • API layer generated from reverse-engineered specs
- • Testing Agent creates test foundation (50%+ test coverage)
Result: New codebase skeleton with tests and docs. Your team can see the shape of the future before building details.
Intelligent Component Extraction (Ongoing)
Extract and rebuild components intelligently. Each iteration faster than the last.
- 1. Isolate: AI identifies component boundaries (code, data, tests)
- 2. Spec: Requirements Agent extracts component specs from legacy code
- 3. Generate: Code Generation Agent rebuilds in modern stack
- 4. Test: Testing Agent validates against legacy behavior
- • Golden path tests: System always behaves like legacy system
- • Compliance Agent validates security/compliance during extraction
- • Documentation Agent auto-generates specs from new code
- • Gradual switchover: New component runs parallel with old
Rhythm: 1-2 week cycles: extract, rebuild, validate, switch. Each cycle, your team learns the new system while legacy still backing up.
Continuous Intelligence & Knowledge Transfer
As you modernize, AI artifacts become institutional knowledge. Your team never forgets why decisions were made.
Auto-generates and maintains docs from code. Every component documented. No manual work.
Architecture decisions recorded. Future engineers see why things were built this way.
Not just watching — actively building new system. You own the result, not us.
Value Delivered
Strategic Outcomes
- Ability to adopt modern tools and integrations
- Reduced hiring difficulty (modern stack attracts talent)
- Foundation for AI-native development
- Reduced security and compliance risk
Typical Engagement
Duration
3-6 months
Development + ongoing advisory option
Investment
CHF 96-288k
CHF 8-12k/week × 12-24 weeks
Team Involvement
2-3 days/week
Your engineers + Fognini team working together
Deliverables
- Architecture assessment report
- Technical debt inventory with prioritisation
- Target architecture documentation
- Refactored components (production-ready)
- Team trained on new patterns