57-Page Analysis in 3 Hours
Using AI to accelerate fractional CTO codebase assessments
57-Page Analysis in 3 Hours
Using AI to accelerate fractional CTO codebase assessments
57-Page Analysis in 3 Hours
Using AI to accelerate fractional CTO codebase assessments
Industry
Industry
Construction Tech (B2B SaaS)
Construction Tech (B2B SaaS)
Company Size
Company Size
10-year-old .NET Framework app.
10-year-old .NET Framework app.
Engagement
Engagement
10,000+ files, massive complexity.
10,000+ files, massive complexity.
Duration
Duration
18+ months (ongoing)
18+ months (ongoing)
Platform
Platform
57-page analysis with executive summary.
57-page analysis with executive summary.
Services
Services
Codebase Assessment
Technical Due Diligence
Architecture Review
Tech Stack
Tech Stack
.NET Framework 4.7.1
Claude AI
GPT-4
Legacy Systems
ASPX
The Challenge
Two years into a digital transformation engagement with a construction tech company, we faced a critical inflection point. The core product was functional, but the path to scaling was unclear. Customer onboarding required extensive custom work, which was destroying unit economics. The ROI on new customer acquisition was not where it needed to be. Before going to the board with recommendations, we needed clarity on what we were actually working with. The codebase was a 10-year-old .NET Framework 4.7.1 application. It had not been maintained to modern standards. Some individual code files exceeded 30,000 lines. The total file count was over 10,000. A traditional CTO assessment of a codebase this size - understanding the architecture, identifying technical debt, mapping dependencies, evaluating modernisation paths - would take 3-4 weeks of focused work. And even then, coverage would be incomplete because human attention cannot scale to that volume of code.
The Challenge
Two years into a digital transformation engagement with a construction tech company, we faced a critical inflection point. The core product was functional, but the path to scaling was unclear. Customer onboarding required extensive custom work, which was destroying unit economics. The ROI on new customer acquisition was not where it needed to be. Before going to the board with recommendations, we needed clarity on what we were actually working with. The codebase was a 10-year-old .NET Framework 4.7.1 application. It had not been maintained to modern standards. Some individual code files exceeded 30,000 lines. The total file count was over 10,000. A traditional CTO assessment of a codebase this size - understanding the architecture, identifying technical debt, mapping dependencies, evaluating modernisation paths - would take 3-4 weeks of focused work. And even then, coverage would be incomplete because human attention cannot scale to that volume of code.
The Challenge
Two years into a digital transformation engagement with a construction tech company, we faced a critical inflection point. The core product was functional, but the path to scaling was unclear. Customer onboarding required extensive custom work, which was destroying unit economics. The ROI on new customer acquisition was not where it needed to be. Before going to the board with recommendations, we needed clarity on what we were actually working with. The codebase was a 10-year-old .NET Framework 4.7.1 application. It had not been maintained to modern standards. Some individual code files exceeded 30,000 lines. The total file count was over 10,000. A traditional CTO assessment of a codebase this size - understanding the architecture, identifying technical debt, mapping dependencies, evaluating modernisation paths - would take 3-4 weeks of focused work. And even then, coverage would be incomplete because human attention cannot scale to that volume of code.
The Approach
Building a Purpose-Built AI Agent
With stability restored, the focus shifted to building a team that could scale without the Fractional CTO being in every conversation.
1. Agent definition
Used GPT to define the requirements for a “modernisation agent” - essentially a prompt engineering exercise to specify what analysis we needed.
2. Agent creation
Implemented the agent definition in Claude, creating a specialised code analysis assistant
3. Iterative analysis
Ran the analysis in stages over 2-3 hours, with the agent processing the entire codebase systematically
The technology stack was old: .NET Framework 4.7.1, HTML files, CSS, ASPX files - patterns that predate modern web development. Claude processed all of it and understood the architectural patterns despite their age.
The Approach
Building a Purpose-Built AI Agent
With stability restored, the focus shifted to building a team that could scale without the Fractional CTO being in every conversation.
1. Agent definition
Used GPT to define the requirements for a “modernisation agent” - essentially a prompt engineering exercise to specify what analysis we needed.
2. Agent creation
Implemented the agent definition in Claude, creating a specialised code analysis assistant
3. Iterative analysis
Ran the analysis in stages over 2-3 hours, with the agent processing the entire codebase systematically
The technology stack was old: .NET Framework 4.7.1, HTML files, CSS, ASPX files - patterns that predate modern web development. Claude processed all of it and understood the architectural patterns despite their age.
The Approach
Building a Purpose-Built AI Agent
With stability restored, the focus shifted to building a team that could scale without the Fractional CTO being in every conversation.
1. Agent definition
Used GPT to define the requirements for a “modernisation agent” - essentially a prompt engineering exercise to specify what analysis we needed.
2. Agent creation
Implemented the agent definition in Claude, creating a specialised code analysis assistant
3. Iterative analysis
Ran the analysis in stages over 2-3 hours, with the agent processing the entire codebase systematically
The technology stack was old: .NET Framework 4.7.1, HTML files, CSS, ASPX files - patterns that predate modern web development. Claude processed all of it and understood the architectural patterns despite their age.
Leadership Pipeline (Months 12-18)
The 57-page output covered
Executive summary - written for non-technical stakeholders (CFO, CEO, CSO)
Architecture overview - how the components interconnect
Technical debt inventory - categorised by severity and remediation cost
Dependency mapping - which modules affect which
Risk assessment - single points of failure, security concerns, scalability blockers
Modernisation pathways - options ranging from incremental refactoring to full rewrite, with cost/benefit analysis for each
Prioritised recommendations - what to fix first and why
The critical differentiator: the executive summary was specifically designed for the board. Not technical jargon, but business impact language. “This module is why onboarding takes 3x longer than it should” rather than “This class has high cyclomatic complexity.”
Leadership Pipeline (Months 12-18)
The 57-page output covered
Executive summary - written for non-technical stakeholders (CFO, CEO, CSO)
Architecture overview - how the components interconnect
Technical debt inventory - categorised by severity and remediation cost
Dependency mapping - which modules affect which
Risk assessment - single points of failure, security concerns, scalability blockers
Modernisation pathways - options ranging from incremental refactoring to full rewrite, with cost/benefit analysis for each
Prioritised recommendations - what to fix first and why
The critical differentiator: the executive summary was specifically designed for the board. Not technical jargon, but business impact language. “This module is why onboarding takes 3x longer than it should” rather than “This class has high cyclomatic complexity.”
Leadership Pipeline (Months 12-18)
The 57-page output covered
Executive summary - written for non-technical stakeholders (CFO, CEO, CSO)
Architecture overview - how the components interconnect
Technical debt inventory - categorised by severity and remediation cost
Dependency mapping - which modules affect which
Risk assessment - single points of failure, security concerns, scalability blockers
Modernisation pathways - options ranging from incremental refactoring to full rewrite, with cost/benefit analysis for each
Prioritised recommendations - what to fix first and why
The critical differentiator: the executive summary was specifically designed for the board. Not technical jargon, but business impact language. “This module is why onboarding takes 3x longer than it should” rather than “This class has high cyclomatic complexity.”
How This Changed the Engagement
With a comprehensive analysis in hand after 3 hours instead of 4 weeks, the Fractional CTO could:
Present to the board immediately with data-backed recommendations
Quantify the cost of inaction - technical debt translated to business impact
Propose specific interventions with realistic timelines and investment requirements
Focus CTO time on strategy rather than code archaeology
How This Changed the Engagement
With a comprehensive analysis in hand after 3 hours instead of 4 weeks, the Fractional CTO could:
Present to the board immediately with data-backed recommendations
Quantify the cost of inaction - technical debt translated to business impact
Propose specific interventions with realistic timelines and investment requirements
Focus CTO time on strategy rather than code archaeology
How This Changed the Engagement
With a comprehensive analysis in hand after 3 hours instead of 4 weeks, the Fractional CTO could:
Present to the board immediately with data-backed recommendations
Quantify the cost of inaction - technical debt translated to business impact
Propose specific interventions with realistic timelines and investment requirements
Focus CTO time on strategy rather than code archaeology
The Results
57-page analysis
57-page analysis
57-page analysis
Completed in 2-3 hours
Completed in 2-3 hours
Completed in 2-3 hours
Avoided
Avoided
Avoided
3-4 weeks of manual work
3-4 weeks of manual work
3-4 weeks of manual work
Executive-ready output
Executive-ready output
Executive-ready output
Suitable for non-technical board presentation
Suitable for non-technical board presentation
Suitable for non-technical board presentation
10,000+ files
10,000+ files
10,000+ files
Analysed comprehensively
Analysed comprehensively
Analysed comprehensively
Complete coverage
Complete coverage
Complete coverage
Rather than sampling-based assessment
Rather than sampling-based assessment
Rather than sampling-based assessment
Immediate strategic clarity
Immediate strategic clarity
Immediate strategic clarity
For transformation planning
For transformation planning
For transformation planning
Key Lessons
AI does not replace CTO judgement - it accelerates it.
AI does not replace CTO judgement - it accelerates it.
AI does not replace CTO judgement - it accelerates it.
The 57-page analysis was a starting point, not a conclusion. The Fractional CTO still needed to interpret findings, prioritise based on business context, and design the transformation programme. But they could do that in days rather than weeks because the discovery work was done.
The 57-page analysis was a starting point, not a conclusion. The Fractional CTO still needed to interpret findings, prioritise based on business context, and design the transformation programme. But they could do that in days rather than weeks because the discovery work was done.
The 57-page analysis was a starting point, not a conclusion. The Fractional CTO still needed to interpret findings, prioritise based on business context, and design the transformation programme. But they could do that in days rather than weeks because the discovery work was done.
Legacy codebases are perfect AI use cases.
Legacy codebases are perfect AI use cases.
Legacy codebases are perfect AI use cases.
Modern, well-structured codebases are easy for humans to understand. Legacy codebases with 30,000-line files are not. AI can process volume that would break human attention spans.
Modern, well-structured codebases are easy for humans to understand. Legacy codebases with 30,000-line files are not. AI can process volume that would break human attention spans.
Modern, well-structured codebases are easy for humans to understand. Legacy codebases with 30,000-line files are not. AI can process volume that would break human attention spans.
Executive communication is a distinct skill.
Executive communication is a distinct skill.
Executive communication is a distinct skill.
Asking the AI to produce a non-technical executive summary was as important as the technical analysis. A CTO’s job is often translation - turning technical reality into business decisions. AI can help with both sides of that translation.
Asking the AI to produce a non-technical executive summary was as important as the technical analysis. A CTO’s job is often translation - turning technical reality into business decisions. AI can help with both sides of that translation.
Asking the AI to produce a non-technical executive summary was as important as the technical analysis. A CTO’s job is often translation - turning technical reality into business decisions. AI can help with both sides of that translation.
Time-to-insight matters.
Time-to-insight matters.
Time-to-insight matters.
If this analysis had taken 4 weeks, the board would have made decisions without it. Speed is not about impatience; it is about catching the decision window.
If this analysis had taken 4 weeks, the board would have made decisions without it. Speed is not about impatience; it is about catching the decision window.
If this analysis had taken 4 weeks, the board would have made decisions without it. Speed is not about impatience; it is about catching the decision window.
The ROI on AI tooling compounds.
The ROI on AI tooling compounds.
The ROI on AI tooling compounds.
This was not a one-time trick. The same approach applies to any new engagement, any legacy assessment, any “what are we actually dealing with” question. The methodology is now part of how we operate.
This was not a one-time trick. The same approach applies to any new engagement, any legacy assessment, any “what are we actually dealing with” question. The methodology is now part of how we operate.
This was not a one-time trick. The same approach applies to any new engagement, any legacy assessment, any “what are we actually dealing with” question. The methodology is now part of how we operate.
The Broader Implication
Fractional CTOs are brought in to make decisions and drive change. The constraint is always time - a fractional engagement means limited hours to create maximum impact.
AI tools like Claude fundamentally change the leverage equation. Work that previously required weeks of manual code review can happen in hours. This means:
Faster time-to-value for client companies
More strategic focus from senior technical leadership
Better-informed decisions because analysis is comprehensive rather than sampled
Higher effective value delivered per fractional engagement hour
We are not replacing technical expertise with AI. We are amplifying technical expertise with AI. The CTO who can effectively use these tools delivers 10x the insight in the same timeframe.
The Broader Implication
Fractional CTOs are brought in to make decisions and drive change. The constraint is always time - a fractional engagement means limited hours to create maximum impact.
AI tools like Claude fundamentally change the leverage equation. Work that previously required weeks of manual code review can happen in hours. This means:
Faster time-to-value for client companies
More strategic focus from senior technical leadership
Better-informed decisions because analysis is comprehensive rather than sampled
Higher effective value delivered per fractional engagement hour
We are not replacing technical expertise with AI. We are amplifying technical expertise with AI. The CTO who can effectively use these tools delivers 10x the insight in the same timeframe.
The Broader Implication
Fractional CTOs are brought in to make decisions and drive change. The constraint is always time - a fractional engagement means limited hours to create maximum impact.
AI tools like Claude fundamentally change the leverage equation. Work that previously required weeks of manual code review can happen in hours. This means:
Faster time-to-value for client companies
More strategic focus from senior technical leadership
Better-informed decisions because analysis is comprehensive rather than sampled
Higher effective value delivered per fractional engagement hour
We are not replacing technical expertise with AI. We are amplifying technical expertise with AI. The CTO who can effectively use these tools delivers 10x the insight in the same timeframe.
Client Context
A construction technology company in the midst of a multi-year digital transformation. This specific use case emerged when strategic decisions required deep technical understanding of a legacy codebase. The AI-powered analysis enabled data-driven board conversations that would otherwise have been delayed by weeks or based on incomplete information. Suitable for: CTOs inheriting legacy systems, technical leaders needing to brief non-technical stakeholders, anyone evaluating AI tools for technical due diligence or code assessment.
Client Context
A construction technology company in the midst of a multi-year digital transformation. This specific use case emerged when strategic decisions required deep technical understanding of a legacy codebase. The AI-powered analysis enabled data-driven board conversations that would otherwise have been delayed by weeks or based on incomplete information. Suitable for: CTOs inheriting legacy systems, technical leaders needing to brief non-technical stakeholders, anyone evaluating AI tools for technical due diligence or code assessment.
Client Context
A construction technology company in the midst of a multi-year digital transformation. This specific use case emerged when strategic decisions required deep technical understanding of a legacy codebase. The AI-powered analysis enabled data-driven board conversations that would otherwise have been delayed by weeks or based on incomplete information. Suitable for: CTOs inheriting legacy systems, technical leaders needing to brief non-technical stakeholders, anyone evaluating AI tools for technical due diligence or code assessment.
See More Case Studies
See how we turn bold ideas into fast, reliable, and scalable digital products.
See More Case Studies
See how we turn bold ideas into fast, reliable, and scalable digital products.
See More Case Studies
See how we turn bold ideas into fast, reliable, and scalable digital products.
Your Next Big Product Starts Here
Work with a team that designs, builds, and ships digital products — fast, scalable, and user-first.
Your Next Big Product Starts Here
Work with a team that designs, builds, and ships digital products — fast, scalable, and user-first.
Your Next Big Product Starts Here
Work with a team that designs, builds, and ships digital products — fast, scalable, and user-first.