In a recent post on LinkedIn, Duolingo Chief Engineering Officer Natalie Glance offered a bold mandate, urging the company to become “AI-first” and use artificial intelligence as the “default starting point” for every task.
What the memo doesn’t make explicitly clear is exactly who AI is meant to be the starting point for. As head of Engineering, Glance may be speaking directly to her team of developers and technologists. But what about digital marketers and HR specialists? Or customer care agents and accountants? Is AI also their starting point?
In a true AI-first organization, AI isn’t just for engineers– it’s for everyone and everything.
AI is the tool that empowers people across all functions to self-solve and self-serve. It’s the means by which non-engineers can build, deploy and enhance the solutions they need when they need them. It’s the foundational technology that makes innovation accessible to all.
The now-public Duolingo memo outlines key hallmarks of an AI-first model, including the importance of sharing knowledge to accelerate impact. Building on that principle, this post offers four actionable steps organizations can take to help teams across the business leverage AI to experiment, innovate, and iterate—independently.
Where we are: The traditional Engineering-first model
Where we’re going: The AI-first organization
Becoming AI-first: 4 best practices to reduce Engineering dependency
1. Redefine the role of Engineering—and rewire the way they work with other teams.
To embrace the shift to AI-first, organizations need every team, including Engineering, to rethink their core identity.
For example, in a traditional model, if an operations manager wanted to automate a spreadsheet, the first step would be to submit a ticket to Engineering. However, in an AI-first model, this person is no longer simply a service requester—they are an active designer, builder and implementer. They can use ChatGPT to write the script and AI to test and debug. With the assistance of an AI-powered tool, they can also implement the code and iterate as needed.
In this model, Engineering’s role must also evolve. Their mandate is no longer to build every internal tool or script, but to design and maintain the platforms that empower others to build. They are not gatekeepers of every process and product, but rather the architects of the platform, focused on ensuring stability, security, and scale.
It’s important to note that reducing the dependency on Engineering does not mean displacement. In fact, quite the opposite is true. It is only with the help of AI that engineers can return to their core purpose: engineering systems, not executing tasks.
2. Invest in an AI-powered execution layer—and empower people to use it.
Behind every AI-first organization is an AI-everything tech stack. This advanced and intuitive execution layer is what will enable non-engineers to build, implement and refine their own solutions.
For example, Replit can be used to translate natural language directly into working applications, removing the friction between business needs and technical implementation. Using this tool, Ops teams can describe what they need in plain language, watch it come to life in code, and deploy or iterate instantly.
Our experts estimate that for every $1 invested in AI tools, organizations could save up to $10 in engineering costs. But while the savings potential is significant, tools alone won’t deliver the value—people will.
Unlocking AI’s value means equipping teams to effectively use tools like Replit and Google Apps Script—while building confidence to adopt what comes next. This AI-mindset won’t be built by technical training on specific tools and programs, but broader culture shifts, like dedicating time for AI experimentation, encouraging knowledge sharing and celebrating success.
3. Leverage AI to optimize processes—over and over again.
For modern organizations, the shift to an AI-first model must also include reimagining processes—not just as a matter of efficiency but also future improvement.
Take resume screening, for example. The legacy model had humans reading and deciding. Today, AI reads, scores, and flags edge cases for human review. Tomorrow, AI will learn from those human overrides to refine its decisions.
The goal of every AI use case is to create processes that get better and better with less and less human intervention. To do this, organizations need to invest in four key areas across every major process:
4. Create AI-powered success metrics—and be ambitious with goals.
In an AI-first organization, independence–people being able to self-solve and self-serve– is one of the strongest indicators of success. Beyond traditional KPIs, companies should begin tracking AI-powered performance metrics such as:
- Percentage of improvements made without Engineering support
- Time from idea to implementation
- Number of activated citizen developers
- Volume of AI-generated process optimization suggestions implemented
It is by setting ambitious goals for teams to build their own solutions, add new features and make improvements that companies can reap the full benefits of this technology and an AI-first model.
AI-first for every team and every task
To become an AI-first organization, leadership must do more than encourage experimentation among engineers—they must drive AI adoption far and wide across the organization. That means setting a new cultural baseline where AI is indeed the default starting point for every team and every task.
Are you on the path to becoming an AI-first organization? Turing Intelligence can help. Contact our expert team today to learn more about how we can help your organization leverage the power of AI across every product, process and person.
James Raybould
Ready to turn AI ambition into outcomes?
Whether you’re exploring GenAI pilots or scaling agentic systems, we’ll help you move fast—with strategy, engineering, and measurable results.