The Shift Toward Custom Software in an AI-Driven World
Autonomous companies are emerging with AI replacing human roles, potentially leading to organizations with far fewer employees while maintaining or increasing output. Research suggests this trend could reshape workforce structures, with 41% of companies globally expected to reduce staff due to AI by 2030.
Economies of Scale & Custom Software
Economies of scale may no longer require massive customer bases, as smaller teams reduce operational complexity. This makes building custom software more feasible, especially when AI lowers development costs by 20-45%.
Jevons paradox applies to software: where cheaper, more efficient building processes lead to more software being created rather than less. AI tools amplify this by boosting developer productivity, though perceptions of gains often exceed reality.
Off-the-shelf (OTS) software faces challenges from customization needs, but providers can adapt by offering headless, composable primitives. This evolution won’t transform every business immediately, but evidence points to a gradual shift accelerating suddenly.
Rise of Autonomous Companies
As AI advances, companies are increasingly automating roles, leading to leaner operations. For instance, firms like Klarna and IBM have signaled workforce reductions through AI integration, allowing them to operate efficiently with fewer people. This reduces the need for large customer volumes to achieve scale, as overhead drops significantly.
Workforce Transformation
The term “autonomous companies” refers to entities that increasingly rely on AI and automation to handle core functions, resulting in significantly reduced human staffing. According to Gartner, by 2028, 40% of services could be AI-augmented, freeing employees for higher-value tasks while transforming business models.
This isn’t mere speculation; in 2025 alone, numerous firms announced AI-related layoffs, with companies like HP, IBM, Amazon, Salesforce, Klarna, and Fiverr explicitly linking workforce cuts to AI efficiencies.
- Klarna: Reduced workforce by over 1,000 in 2022 as part of an AI shift.
- UPS: Laid off 20,000 in early 2025, citing automation.
Interestingly, SMBs are particularly resilient. Data from 2025 shows that companies with 500 or fewer employees largely avoided mass AI layoffs, instead using AI as a “force multiplier” to enhance operations without slashing staff.
Software Development: From Bottleneck to Solved Problem?
A core assumption in this shift is that software engineering is becoming “somewhat solved”—not perfectly, but sufficiently repeatable and affordable.
- 85% of developers used AI for coding in 2025.
- 62% rely on assistants that save at least an hour weekly.
- 10-15% productivity boosts reported by Bain.
- 20-45% reduction in development time estimated by McKinsey.
The Perception-Reality Gap
A 2025 randomized trial by METR found developers using early-2025 AI tools took 19% longer on tasks, despite perceiving a 20% speedup. This highlights challenges like prompt engineering and context management. However, trends point upward: Graphite notes 30-50% faster task completion with AI, reducing bugs by 40% in some teams.
With 30 million developers worldwide, AI could double productivity, adding trillions to GDP.
Jevons Paradox in Software
As efficiency rises, consumption increases. In software, cheaper building leads to more creation, not less. Historical examples include coal engines and fuel-efficient cars; today, AI tools explode demand for custom solutions. As one analysis notes, “every leap in efficiency creates more developers, not fewer.”
Viability of Custom Software at Lower Scales
Building operational software—e.g., for comms or support—is often trivial but scales poorly with customer growth. AI changes this by enabling economies of scale with fewer customers. Custom builds become viable, offering tuning without OTS compromises.
High-revenue, low-headcount companies exemplify this:
- NVIDIA: Generates $4.4 million per employee with 29,600 staff.
- VICI Properties: Tops lists at $142 million per employee (27 staff).
- Netflix: Hits $4.15 million per employee.
The Future of OTS: Primitives and Composability
Off-the-shelf software won’t vanish, but it must evolve. Providers should offer primitives—modular, API-first components—for easy integration.
- Headless Commerce: Platforms like CommerceTools enable composable stacks.
- Malleable Software: OTS that’s user-customizable via AI, blending build and buy.
By 2026, AI budgets are expected to rise 36%, favoring custom for cost-efficiency.
Sector-Specific Impacts
| Sector | OTS Reliance | Custom Shift Potential | Examples |
|---|---|---|---|
| Tech | High (SaaS) | Very High (AI coding) | NVIDIA ($4.4M/emp) |
| Energy | Medium | High (Automation) | Cheniere ($15.7B rev) |
| Finance | High | Medium (Fraud AI) | Klarna (AI cuts) |
| Retail | High | Growing (Composable) | Amazon (Workforce) |
Conclusion
AI’s role in fostering autonomous companies and democratizing software creation heralds a custom-centric future. While OTS adapts via composability, the balance tips toward tailored solutions, driven by efficiency paradoxes and economic realities.
Key Citations
- Companies Are Replacing Workers With AI—Fast (Forbes)
- 41% of Companies Expect to Replace Workers With AI (Vice)
- Klarna’s AI customer support does the work of 700 people
- Ranked: The Top Companies via Revenue Per Employee (Visual Capitalist)
- The Jevons Paradox and AI Efficiency (Proxify)
- Composable vs. Headless Commerce Explained (Contentful)