Dear CIO,

Today, we're diving into a topic that’s reshaping software engineering in the AI era. As artificial intelligence tools revolutionize workflows, CIOs and technology leaders must navigate a balance between simplifying processes for efficiency and enriching them with the contextual depth that drives innovation. Partially inspired by Patrick Debois’ recent presentation, “AI Product Engineering: how AI is changing our products” (see link below), this edition explores the concept of “collapsing the stack” to streamline development while simultaneously “expanding the stack” with robust engineering practices that maximize AI's potential. We’ll uncover how to strike the right balance between simplicity and substance in this brave new world of AI-native development.

Best Regards,
John, Your Enterprise AI Advisor

The AIE Network is a network of over 250,000 business professionals who are learning and thriving with Generative AI, our network extends beyond the AI CIO to Artificially Intelligence Enterprise for AI and business strategy, AI Tangle, for a twice a week update on AI news, The AI Marketing Advantage, and The AIOS for busy professionals who are looking to learn

From Code to Context

Transforming Software Engineering in the AI Era

AI tools have sparked a seismic shift in software engineering, pushing us to rethink traditional workflows and embrace a new way of thinking. While much of the conversation focuses on collapsing the stack to reduce complexity, the real opportunity lies in simultaneously expanding the stack to include known software engineering practices that AI tooling can thrive on—in other words, not just code. This duality—collapsing the stack while enriching its context—is key to leveraging the full potential of AI-native development.

From Complexity to Efficiency: The Case for Collapsing the Stack

Building applications no longer requires massive teams of data scientists or complex, custom models to use AI. Today’s AI engineers can access platforms that condense intricate processes into simple, deployable solutions. By collapsing the stack, we reduce friction, minimize technical debt, and enable engineers to iterate faster. However, this streamlined approach introduces a new challenge: how do we ensure that simplification doesn’t come at the expense of quality?

The answer lies in expanding the stack—not in terms of complexity, but in the context we provide.

Expanding the Stack with Context: Best Practices as AI Engineering Expands

Tools like GitHub Copilot and Cursor demonstrate a fundamental truth: the better the input, the better the output. These tools enable software engineers to pull from a vast array of contextual information—test coverage, API documentation, designs, requirements, incidents, and runbooks—that can generate valuable insights. In doing so, they can elevate the importance of good software engineering practices, often overlooked as tedious or secondary.

Interestingly, AI engineering might warn us that to go faster, we need to slow down and get our house in order. Standardized processes, comprehensive documentation, and well-maintained knowledge repositories aren’t just good practices; they’re enablers of AI-driven productivity. Feeding AI tools with high-quality context ensures they can deliver on their promise, making these practices essential rather than optional. One could argue that we now have no choice with AI’s nondeterministic approach.

The Shifting Role of AI Engineers

In this AI-powered landscape, engineers' roles are evolving. Traditional distinctions like front-end, back-end, or database engineering give way to a customer-centric focus. Engineers are no longer just builders but problem-solvers, refining applications to meet user needs and expectations. 

The success of an AI-created service hinges on its ability to deliver consistent, reliable outcomes. Engineers must go beyond ensuring functionality; they must now optimize user experience, address edge cases, and fine-tune their solutions. In short, they’re moving from building apps to optimizing them, a shift made possible by collapsing and expanding the stack simultaneously.

Lessons from Digital Transformation: Tools Alone Are Not Enough

The digital transformations of the past decade taught us that tools can only take us so far. Real impact comes from rethinking processes, standardizing workflows, and integrating best practices. AI engineering is no different. To make AI tools shine, organizations must adopt a holistic approach, ensuring that their processes and practices align with the needs of AI-native workflows.

Knowledge management is central to this. Artifacts like test coverage, documentation, and runbooks must be updated and integrated into the development process. These same AI tools must enable context awareness of performance, security, and business requirements as part of the design requirements. AI tools can operate at their full potential when this context is readily available, transforming tedious tasks into seamless processes.

Simplifying Access, Amplifying Impact

Engineers need powerful but accessible tools to keep up with the ever-evolving AI landscape. Platforms with low learning curves that include products like MongoDB, Kafka, Kubernetes, and many other extensions must provide a solid foundation for engineers of all experience levels. These tools reduce barriers to entry while offering the flexibility to address industry-specific challenges.

The goal is to remove friction, enabling engineers to focus on solving real problems. By collapsing the stack to simplify workflows and expanding it to include a good software engineering context, organizations can create a seamless, empowering environment for their engineering teams.

Looking Ahead: The Future of AI Engineering

The future of AI engineering lies in balancing complexity and impact. Collapsing the stack makes development faster and more efficient, but expanding the stack ensures that the output is valuable, reliable, and impactful. This dual approach allows us to integrate context-rich best practices into default workflows, making them the norm rather than the exception.

As we embrace AI-native development, the lesson is clear: success doesn’t come from tools alone but from how we integrate them into a cohesive, well-informed process. By collapsing the stack and expanding its context, we can create new possibilities for innovation and efficiency in software engineering.

DevOps promised us a new era where speed and quality can complement rather than compete. We are beginning a new software engineering era where, as my good friend Andrew Clay Shafer says, we should make the right thing the easy thing. The “stack” is ready for a rethink. In this “AI Engineer” era, the key lies in striking the right balance between simplicity and substance.

Patrick’s Presentation

• Frank Wang explores the balance between applying foundational security principles like data protection, API security, and addressing novel AI-specific risks.

• Michael Bruemmer and Jim Steven look at Experian’s 12th annual Data Breach Industry Forecast which highlights five major cybersecurity trends for 2025.

• Andre Quintanilha covers LangChain’s research reveals widespread plans for AI agent adoption but highlights challenges in explainability, real-time monitoring, and the limitations of legacy observability tools.

• A research report from Ivanti shows how organizations are using Gen AI and cybersecurity, and the risks that accompany them.

• The FBI released a PSA on how criminals are using Gen AI for financial fraud.

• Arize has launched Phoenix 6.0, introducing the Prompt Playground—a feature that streamlines prompt engineering by enabling users to test and compare prompts, and more.

• David Linthicum describes how companies market their AI models as "open" to align with transparency and collaboration ideals, while keeping critical components proprietary.

• Ankit Bargale announces IBM’s QRadar Generative AI Content Extension, available for free on IBM App Exchange, enabling Security Operation Centers to monitor the usage of popular generative AI tools.

• Jérémy Ravenel analyzes John Sowa's talk, "Without Ontology LLMs are Clueless," which argues that human thought transcends language, emphasizing the need for ontologies in AI to bridge the gap between mental models and linguistic processing.

• The All Things Open AI Conference will take place in Durham, NC on March 17-18.