Enterprise AI Analysis
What do professional software developers need to know to succeed in an age of Artificial Intelligence?
Authors: Matthew Kam, Google; Cody Miller, Google; Miaoxin Wang, Trilyon; Abey Tidwell, Google; Irene A. Lee, Education Development Center; Joyce Malyn-Smith, Education Development Center; Beatriz Perret, Boston College; Vikram Tiwari, Assembled; Joshua Kenitzer, Google; Andrew Macvean, Google; Erin Barrar, Google.
Abstract: Generative Al is showing early evidence of productivity gains for software developers, but concerns persist regarding workforce dis-ruption and deskilling. We describe our research with 21 developers at the cutting edge of using AI, summarizing 12 of their work goals we uncovered, together with 75 associated tasks and the skills & knowledge for each, illustrating how developers use AI at work. From all of these, we distilled our findings in the form of 5 insights. We found that the skills & knowledge to be a successful AI-enhanced developer are organized into four domains (using Generative AI effectively, core software engineering, adjacent engineering, and adjacent non-engineering) deployed at critical junctures through-out a 6-step task workflow. In order to "future proof" developers for this age of AI, on-the-job learning initiatives and computer sci-ence degree programs will need to target both "soft" skills and the technical skills & knowledge in all four domains to reskill, upskill and safeguard against deskilling.
Executive Impact & Business Value
Our research with 21 leading AI-enhanced software developers reveals profound shifts in productivity and required skills.
0
of professional developers already use AI tools
0
productivity improvement (Pull Requests completed weekly)
0
tasks identified where GenAI is used
0
skills & knowledge domains for success
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Domain 1: Generative AI Usage
This domain revolves around effective engagement with AI tools, including being able to assess AI's outputs and calibrating interactions with LLMs to achieve desired outcomes (see section 4.3 for more details). This domain also includes foundational knowledge of machine learning; as well as AI tools' capability and constraints (e.g., AI failures, including how and why AI hallucinates), which enable developers to select appropriate tools for given tasks and avoid wasting time using AI due to unrealistic expectations about AI.
Domain 2: Core Software Engineering
This encompasses the essential skills and knowledge for software development, which remain crucial for making sound technical decisions in the age of GenAI. Three key aspects about this domain include 1) coding and testing proficiency - knowing what constitutes high-quality maintainable code, best practices in defensive coding, optimization, systematic testing approaches, and debugging skills with comprehension and critical thinking; 2) risk assessment for production readiness (e.g., being able to recognize potential technical compro-mises, vulnerabilities, and planning for contingencies); and 3) good system design (e.g., possessing a deep understanding of existing and alternative system architectures, design & system constraints, and carefully weighing the potential benefits and drawbacks of multiple design solutions to meet requirements).
Domain 3: Adjacent Software Engineering
Specialized sub-domains within and closely related to software engineering e.g. cybersecurity, industry-specific regulations, and emerging techno-logical trends. The exact topics depend on the developer's context.
Domain 4: Adjacent Non-Software Engineering
This encompasses at least 5 distinct sub-domains as we learn from the profile: end-user, customer, business or industry, competitors landscape, and market trends. Similar to Domain 3, the specific topics (e.g. General Data Protection Regulation) depend on the organiza-tional context. For example, developers in smaller companies may need the skills & knowledge to "translate product (and technical) requirements into business needs (and vice versa)", "analyze feed-back from users and internal stakeholders" and "integrate needs and feedback into work item". Developers in larger companies may focus more on contextualizing their work items so they are bet-ter able to evaluate trade-offs (e.g., cost-benefit analysis) and risks when optimizing for business value.
63%
of professional developers already use AI tools
Enterprise Process Flow (6-step Workflow)
Identify
→
Engage
→
Evaluate
→
Calibrate
→
Tweak
→
Finalize
Insight #1: AI for Optimization
With less time spent on repetitive work, AI frees developers to experiment - under shorter cycles and come up with optimal solutions to problems. From the profile, we observe that developers seek to use AI to optimize for both business metrics (e.g. "selects one system that is most vetted, most efficient, cost effective and has better chance of success and profitability") and engineering metrics (e.g. "complexity, performance, maintenance").
Insight #2: AI for Better Documentation
In some situations, AI tools facilitate individual developers in meeting their team's greater good. For example, generating example code for tutorial documentation (task) when producing up-to-date documentation (goal) can be time-consuming in the short term. With AI, developers complete this task more efficiently, and potentially reduce disruptions and save even more time in the long term for themselves and coworkers - when coworkers can find answers from better documentation instead of having to ask.
Insight #3: Collaboration Complexity
While early research indicates that AI's time savings promote collaboration, our research suggests that the reality is more complex. When AI helps with finding previous code authors and explaining code, there is a reduced need for direct communication. Panelists report spending less time communicating with others. Similarly, an advisor notes that unless AI is used with care (e.g. AI-generated summaries of emails), human interactions can be reduced to information exchanges devoid of the human touch. On the other hand, when AI saves time asking and answering simpler questions, AI creates the potential for human interactions to focus on more meaningful questions. The same advisor adds that AI can help to put people in touch: "AI more like a 'trusted advisor' rather than an 'executive' would probably be more of a net positive: less 'Here is a summary of your coworkers doc' and more 'Hey @ldap, we think this doc written by @foo relates to your existing work and plans."
Insight #4: AI Reshapes Content Knowledge Acquisition
With the rise of AI, developers are changing how they acquire content knowledge. As one advisor puts it: "Before GenAI... I'd have to maybe read books [respondent's emphasis]... While I [still] need to be familiar with other [programming] languages, [this] can be an issue for junior developers because I acquired this [knowledge]... by ... reading."
Insight #5: The Need for Holistic Skills & Workflow Understanding
Most importantly, the above description of the workflow shows where in the workflow, and how, Domains 2, 3 & 4 are absolutely critical for developers to effectively leverage AI tools. This perspective goes beyond prompt engineering in presenting a holistic understanding of the skills & knowledge that are critical for AI-enhanced software development. In fact, it calls for fluency in these skills & knowledge in order for the developer as a user of AI to be effective as the human in the loop, exercising appropriate control over AI throughout the workflow.
Advanced ROI Calculator
Estimate the potential productivity gains and cost savings from AI implementation in your engineering organization.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Your Enterprise AI Implementation Roadmap
A structured approach to integrate AI into your software development workflow, ensuring sustainable growth and optimal performance.
Phase 1: Assessment & Strategy
Understand current workflows, identify AI integration points, and define strategic objectives.
Phase 2: Pilot & Experimentation
Implement AI tools in a controlled environment, gather feedback, and iterate on best practices.
Phase 3: Skill Development & Training
Upskill engineering teams on AI tools, prompt engineering, and critical evaluation of AI outputs.
Phase 4: Phased Rollout & Integration
Gradually expand AI adoption across teams, integrating tools into existing CI/CD pipelines.
Phase 5: Monitoring & Optimization
Continuously track AI performance, gather metrics, and fine-tune models for maximum impact.
Phase 6: Future-Proofing & Innovation
Stay ahead of AI advancements, explore new capabilities, and foster a culture of AI-driven innovation.
Ready to Transform Your Engineering Practice with AI?
Unlock unparalleled productivity, reduce development costs, and empower your team with the future of software engineering.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat