Software 3.0: How AI and Composability Are Reshaping Enterprise Tech
Enterprise software is evolving from AI-driven composable systems, requiring organizations to reshape both their tech strategies and leadership models.
Speaking at Y Combinator’s AI Startup School in June 2025, OpenAI co-founder Andrej Karpathy stated, “Software has not changed much on such a fundamental level for 70 years. And then I think it’s changed about twice – quite rapidly – in the last few years.”
His observation underscored a pivotal truth: the ground is shifting beneath every developer, user, and business leader. And the way we design and deliver software will never be the same.
Welcome to Software 3.0
For decades, software served as the invisible infrastructure on which modern organizations were built. We structured teams around it, codified workflows through it, and grew entire functions around specific systems – marketing lived on CRM platforms, finance teams on ERP systems, and supply chains in planning suites.
That legacy approach created a familiar rhythm: vendors built monoliths, companies adapted their processes to fit, and upgrades came on multi-year cycles. CIOs were custodians, while business leaders remained consumers. Software was treated like heavy infrastructure – a CapEx bet with ROI measured in years.
But now we’ve entered a fundamentally different era. We’ve evolved from Software 1.0 (writing code) and Software 2.0 (training neural networks) to Software 3.0, where natural language becomes the programming interface. Large AI models automatically generate code, wire together APIs, test and deploy workflows, and even optimize performance in real time. What once required armies of developers and months of work can now be composed in hours.
Driven by breakthroughs in AI, agentic workflows, low-code/no-code, the redefinition of the tech stack, and a wave of composable middleware, the very fabric of software is being rewoven – with profound consequences.
The Complexity Crisis
For years, every new requirement produced the same reflex:
- Add a workflow to manage it
- Create a hierarchy layer to control it
- Insert another technology layer to bridge gaps
- Bring in another provider to supply it
Each decision seemed rational in isolation. But over time, they created an exponential “Complexity Index” – more handoffs, more integrations, slower decisions, and higher costs.
The numbers speak for themselves: Zero100 has found that a typical Fortune 500 supply chain now has at least 1,400 enterprise-level software providers to choose from – nearly triple the number available five years ago.
Meanwhile, tech stacks have added countless hidden layers – cloud services, open-source libraries, and embedded AI models. Integration complexity has increased by at least 4x in just a few years.
A New Leadership Model
In this era of composability and agentic workflows, the biggest design decisions aren’t about code – they’re about workflows.
Three leaders must now act together:
- The Technologist designs the ecosystem
- The COO/CSCO tears down barriers and orchestrates end-to-end processes
- The Functional VP creates domain excellence
Technology and operations are converging in real time, and functional boundaries are dissolving. Operations leaders need to be in the room shaping those choices.
Zero100 data shows that supply chain organizations are already preparing for this convergence, hiring for roles that blend AI, data science, and workflow design, such as semantic search engineers and multi-agent system engineers. Demand for certain roles has risen steadily since 2023, with some spiking to nearly 1% of workforce share.
The Composability Advantage
Composability offers a way forward, creating swappable layers where capabilities can be added, removed, or upgraded without tearing down the whole stack. Think of it as a way to design for the trajectory rather than the snapshot of where you are today.
As venture capitalist Chris Dixon put it, “Composability is to software as compounding interest is to finance” – the benefits accumulate and accelerate over time.
This doesn’t mean the old platforms are dead. Far from it. What’s changing are the expectations: platforms must interconnect, expose data as products, and allow new tools to dock in without friction. Composability lets organizations wrap new capabilities around existing systems, extending their value rather than replacing them entirely.
The future is hybrid, combining the scale and stability of established platforms with the agility of modular extensions. The best leaders will resist the false choice of “build vs buy,” adopting a “build and buy” mindset where strategic advantage sits on top of commercial infrastructure.
Shell demonstrates this at scale. The company has built and owns more than 100 forecasting models running on an open, plug-and-play ingestion and data mesh platform. Shell can swap data sources and deploy new models without rewiring its core systems – achieving both control and flexibility.
Zero100 research confirms the advantage: our data shows that organizations with higher adoption of data products and composable architectures outpace others in terms of revenue (+132 basis points, 3-year compound annual growth rate) with less capital investment (-30 basis points).
Why This Matters Now
Software is shifting toward composable, AI-driven workflows. Vendors are exposing APIs and connectors. Early adopters are already seeing faster response times and lower integration costs.
The leadership imperative is clear: traditional signoffs, budget cycles, and governance were built for quarterly cadence and multi-year rollouts, but today’s models move at weekly speed.
Success requires adjusting both who participates in technology decisions and the principles used to evaluate them. Focus on openness, workflow impact, time-to-value, and interoperability. This approach will help you capture the benefits of this new era and keep your operations adaptive and fast.
The question isn’t whether this transformation will happen – it’s whether you build for today or build for the change.