Felmo's AI-Driven Engineering Efficiency Strategy
An executive analysis of how Felmo leverages GenAI to reduce headcount while maintaining output. The CTO details the shift from Cursor to Claude Code for cost efficiency, the elimination of manual QA loops, and the strategic focus on ROI-driven product development in a tech-enabled service model.
Executive Summary
Felmo, a tech-enabled mobile veterinary service, demonstrates a pragmatic approach to integrating Generative AI into its engineering and operational workflows. Unlike pure tech startups, Felmo leverages technology to optimize a service-heavy business model, where efficiency directly impacts unit economics. The CTO, Stefan Schubert-Peters, outlines a strategy focused on cost efficiency, headcount optimization, and rapid iteration, providing a clear blueprint for startups navigating the AI transition.
Strategic Shifts in Tooling and Cost
A key strategic decision was the migration from Cursor to Claude Code subscriptions. This shift was driven by economic constraints rather than technical superiority. By analyzing token usage and subscription costs, Felmo determined that direct model subscriptions offered 2-3x more value than API-based tools. This move highlights the importance of treating AI tooling as a variable cost that must be optimized for ROI, not just a productivity enhancer.
Operational Efficiency and Headcount
The impact of AI on headcount is tangible. Felmo is not backfilling departed roles, citing AI-driven productivity gains that allow the remaining team to handle increased workloads. Approximately 50% of code pushes are AI-generated, and manual QA loops are being eliminated in favor of automated, AI-assisted testing. This reduction in human-in-the-loop processes accelerates deployment cycles and reduces the cognitive load on developers, allowing them to focus on high-level architecture and business logic.
Governance and Quality Control
Governance at Felmo is enforced through non-functional requirements (NFRs) such as load times and test coverage, rather than rigid procedural controls. AI tools assist in meeting these metrics, ensuring that quality standards are maintained without slowing down development. This approach allows for a flexible, agile environment where developers can experiment and iterate quickly, provided they meet the core performance benchmarks.
Market Implications
The adoption of AI is raising the baseline for software quality across the industry. As 'average' standards rise, differentiation must come from deep customer understanding and superior business outcomes. Felmo's strategy focuses on using AI to enhance the core service experience, such as automating documentation and recommendations for veterinarians, rather than building new products. This focus on core business optimization ensures that AI investments directly contribute to revenue and customer satisfaction.
Conclusion
Felmo's approach offers a replicable model for startups looking to integrate AI effectively. By focusing on cost efficiency, headcount optimization, and strict quality metrics, the company has achieved significant productivity gains without compromising on quality. This strategy emphasizes the importance of aligning AI adoption with business goals, ensuring that technology serves the core mission rather than becoming a distraction.
Key insights
-
Economic constraints drive tool selection more than feature sets. Felmo switched to Claude Code because it offered better value per dollar than Cursor, despite Cursor having preferred features like model independence.
Impact: Startups can significantly reduce AI spending by choosing subscription models over API-based tools, freeing up budget for other strategic initiatives.
-
AI-generated code now constitutes 50% of all pushes, with a third of developers primarily using AI for coding. This shift has allowed the company to maintain high velocity without increasing headcount.
Impact: Companies can scale output without scaling headcount, improving margins and allowing for more agile organizational structures.
-
Manual QA loops are being eliminated in favor of automated, AI-assisted testing. This reduces bottleneck time and allows developers to focus on higher-value tasks like specification and review.
Impact: Reducing manual QA accelerates deployment cycles and improves developer satisfaction by removing repetitive, low-value tasks.
-
Governance is enforced through non-functional requirements (NFRs) like load times and test coverage, rather than rigid procedural controls. AI tools assist in meeting these metrics, ensuring quality without slowing down development.
Impact: A metrics-based governance approach allows for greater flexibility and agility, while still maintaining high quality standards.
-
AI adoption is raising the baseline for software quality across the industry. As 'average' standards rise, differentiation must come from deep customer understanding and superior business outcomes.
Impact: Companies must focus on customer-centric innovation to stand out in a market where technical quality is becoming a commodity.
Action items
-
Audit current AI tooling costs and compare subscription vs. API pricing to identify potential savings. Consider switching to direct model subscriptions if they offer better value.
Impact: Reduces operational costs and frees up budget for other strategic initiatives, improving overall financial health.
-
Implement automated, AI-assisted testing to reduce reliance on manual QA. Focus on high-risk areas for manual review, but automate standard checks.
Impact: Accelerates deployment cycles and improves developer satisfaction by removing repetitive, low-value tasks.
-
Define and enforce non-functional requirements (NFRs) such as load times and test coverage as the primary governance mechanism. Use AI tools to help meet these metrics.
Impact: Ensures quality standards are maintained without slowing down development, allowing for greater flexibility and agility.
-
Evaluate the impact of AI on headcount and consider not backfilling departed roles if productivity gains justify it. Monitor the effect on team morale and workload.
Impact: Improves margins and allows for more agile organizational structures, while preserving budget headroom for innovation.
-
Focus AI investments on enhancing the core business experience rather than building new products. Ensure that AI tools directly contribute to revenue and customer satisfaction.
Impact: Ensures that AI adoption is aligned with business goals, maximizing ROI and driving sustainable growth.
Quotes
“Ich würde behaupten, dass die Hälfte unseres Codes, also die, sagen wir mal, die Hälfte unserer Pushes AI-generiert sind.”
“Wir machen es aber typischerweise erst zum Thema, wenn es wirklich ein handfestes Problem gibt.”
“Ich glaube, der Schlüssel wird dann sein, wie stark diese Entwickler tatsächlich an Marktbedürfnissen, an Kundenbedürfnissen dran sind.”