Want to transform with AI? Emulate your engineers

Software development is ground zero for AI transformation. 

While boardrooms draw up AI strategies, engineers are already taking advantage of the perfect laboratory conditions for shifting work from the inside out. And their “lab” reveals a stark fact: The gap between early movers and stragglers isn’t narrowing — it’s widening. In fact, enterprises leading AI adoption say they’ve already seen a 37% lift in productivity. 

Key takeaways:

• Software engineering provides the ideal environment for AI agent development with instant validation cycles, ruthless tool selection, and constant innovation
• Engineers create a perpetual improvement flywheel as they simultaneously build and use AI tools, experiencing pain points and benefits firsthand
• Design AI-native systems rather than bolting AI onto existing applications to achieve meaningful productivity gains
• Teams must master effective prompting and output review skills to successfully implement AI agents across knowledge work domains
• Companies that fundamentally redesign workflows around AI agents will outpace those settling for incremental improvements in execution velocity

All eyes on engineers

Coders are basically their own users. With a value cycle that’s entirely digital and nearly self-contained, they can assess value in minutes instead of months. They’re motivated to identify what works (and they’ll drop what doesn’t). 

Plus — with AI having transformed coding workflows so completely that frameworks quickly become obsolete — they’re accustomed to fast pivots. They’re also capitalizing on an innovative landscape where, according to the State of AI report, “AI enables faster product development, real-time customer feedback analysis, and predictive trend forecasting.” Box CEO Aaron Levie said, "The very builders of the systems know the workflows, so they can improve the tech faster. And they have a huge incentive to make their tools better to create a perpetual flywheel." 

This combination — builders as users, instant feedback, and willingness to pivot — creates the perfect conditions for AI agents to evolve at breakneck speed. It’s why coding represents the leading edge of the AI agent revolution.

Engineering skillsets everyone can adopt

Want to overhaul workflows like an engineer? If you’re a business that deals with digital information, you can achieve considerable success. 

"While coding is ahead of the curve right now in agentic workflows, it's clear this pattern will start to emerge across most other domains over time,” Levie explained. “Leverage is going to show up everywhere: the ability to generate 10x more marketing output, process deals way faster due to automated legal workflows, handle customer support and success operations in faster ways, and so on.”

Consider best practices like designing natively for AI (rather than bolting innovations onto systems) and reorganizing processes around agents.

“If you want to realize the promise of AI, especially in the workflow automation space, you cannot just take your AI and add it on top of your applications,” Nirmal Ganesh, Senior Team Director, Product Management, explained in a recent workflow automation discussion. “... You have to bake AI in the core experience, and that's where you will see productivity gains. If you application is not AI native, you will not be able to see the productivity gains that you expect."

Specific practices that raise execution velocity:

• Designing hyper-specific prompts and iterating until agents perform reliably
• Running many agents in parallel to explore solutions and speed up execution
• Prioritizing reviewover manual work, treating agent output as raw material to be tested and integrated

Most importantly, don’t stop at incremental change. Good engineering builds for scalability and resilience. 

“Most existing teams and companies will be happy about incremental gains and stop short of doing the really big changes in how they work,” Levie noted. “And doing so is going to provide at least a temporary advantage to those that adapt to these new ways of working.”

Across departments, changes can add up to improvements like greater marketing output, automated legal workflows that speed up deals, and faster customer support operations. It’s a future where, according to Box research, knowledge workers effectively become “managers of AI Agents… clearly defining tasks and desired outcomes — as well as quality-checking agent work.” 

Here’s how an AI-first redesign might show up: 

• Teams instantly analyze customer surveys with natural language processing (NLP), which categorizes feedback without manual review
• Content automatically becomes insights as data gets extracted from contracts, invoices, and other documents
• Processes are automated, which means faster turnaround on tasks like employee onboarding and training
• Due diligence happens faster, even across thousands of M&A transaction documents
• Trends are instantly identified and put to work across product research data

Practical ways to get started 

Applying software design principles across departments might sound like a lofty goal, but it’s manageable when you start with small, concrete changes to how teams approach their daily work. 

As your first easy win, you can get workflows moving in the background by building knowledge (more in this guide) around two key engineering skills:

1. Prompting agents effectively: Framing tasks so agents produce useful outputs
2. Reviewing agent outputs: Validating what your agent produces and then iterating

Next, identify a workflow where AI agents can fundamentally alter outcomes — and redesign it. Choose a process with clear inputs and outputs, immediate feedback potential, and high repetition value. 

In practice, that looks like:

• Recreating your process so it’s AI-native and automation is embedded
• Running rapid feedback loops and treating users as builders who immediately test and refine outcomes
• Defining tasks with precision and creating acceptance criteria, so agents (and people) produce verifiable results
• Shifting your focus from manual execution to review and orchestration, emphasizing quality-checks instead of manual work
• Adopting practices in continuous improvement that favor breakthrough changes over incremental tweaks

The blueprint’s in the code

Engineers have demonstrated that AI agents have shifted software engineering in profound ways, and that transformation is only beginning. The changes emerging today will inform how work gets done across knowledge industries for years to come. 

“My hunch is that many of the best practices that are emerging in coding agents will eventually transfer to other categories of AI agents across all of knowledge work,” Levie said. “The tools will have to get even simpler, but the paradigms will persist. It’s clear that almost everything about software development has fundamentally changed for good with AI agents.”