AI Needs to be Compulsory

Ray Poynter, 13 August 2025

Beware of the Good, the Bad and the Ugly of mandating AI adoption.

There is widespread agreement amongst seasoned consultants that if an organisation is going to realise the benefits of AI fully, it needs to be adopted throughout the organisation. I firmly believe this to be true, and I am working with a number of organisations to help them achieve this. However, there are good ways and bad ways of going about this that I will illustrate with three case studies. Norges Bank, Shopify, and Klarna.

Norges Bank Investment Management (NBIM) – The Prototyping Mandate

The world’s largest sovereign wealth fund, Norges Bank Investment Management (NBIM), provides a model for a successful, hybrid approach to AI adoption. The mandate, led by CEO Nicolai Tangen, is unequivocally clear: using AI is a “must,” not an option. This top-down vision is a central component of the fund’s three-year strategic plan, but its implementation is far from a rigid, one-size-fits-all dictate.

NBIM’s strategy pairs this clear executive vision with a robust, decentralized, and experimental culture. The firm’s ethos is to “build our way forward,” and it encourages departments and investment units to conduct their own AI projects. This grassroots innovation is supported by a network of “AI ambassadors”—volunteers who spread knowledge about the technology and help their colleagues “unlock the door” to its potential. This creates a critical feedback loop, allowing the firm to prototype and test new tools in a risk-controlled environment. A key use case is an internally developed AI engine that monitors and measures portfolio managers’ skills to identify and correct behavioural biases, such as herd behaviour. This promotes a culture of “psychological safety” where managers are encouraged to be contrarian, ultimately leading to better decision-making. The success of this approach is already evident, with NBIM reporting that an AI tool is saving an initial $100 million in annual trading costs, with an ultimate target of $400 million.

Shopify – The “AI-First” Cultural Shift

Shopify’s CEO, Tobi Lütke, has been a vocal proponent of a more aggressive, cultural mandate. His internal memo, later made public, declared that “reflexive AI usage is now a baseline expectation” for every employee. The memo’s intent was to move AI from a mere suggestion to a non-negotiable professional skill, comparable to email proficiency.

The mandate is designed to change the way work is done at the company fundamentally. It formalises AI usage as an explicit metric in performance reviews, integrates AI into all prototyping phases, and requires teams to justify why a task cannot be done with AI before requesting more headcount. This strategy aims to foster a workforce of “super developers” and “AI-prolific” workers, who can be 10 times or even 100 times more productive. While ambitious, this approach has also drawn criticism. Observers caution that “reflex without reflection” is dangerous, as AI is a multiplier that can just as easily amplify biases, laziness, and poorly defined problems. The focus on efficiency as the only valid metric, critics argue, risks devaluing slow, thoughtful, or human-centric work and could lead to output without genuine insight.

Klarna – The Cautionary Tale of Over-Automation

The Swedish fintech company Klarna offers a compelling case study on the risks of a purely cost-driven, top-down mandate. Klarna’s CEO, Sebastian Siemiatkowski, was an outspoken advocate for AI as a “workforce replacement,” aggressively automating the work of 700 customer service agents. The company’s initial strategy was to leverage AI for speed and cost savings, believing the technology could match or exceed human performance in a variety of tasks.

This aggressive, efficiency-focused approach ultimately backfired. Klarna experienced a surge in customer complaints and a steep decline in user satisfaction, which forced a public reversal of the policy. Siemiatkowski later admitted that focusing too heavily on cost had led to “lower quality,” acknowledging that the “human element in service roles remains indispensable”. The company’s subsequent strategic shift—re-hiring human agents, offering a human-led customer service option for VIP clients, and launching a new “AI CEO hotline” for feedback—reveals a more nuanced, hybrid model where AI serves to enhance human interaction rather than replace it. This experience underscores a crucial lesson: while AI can be highly efficient, it is not always efficacious in delivering the quality and trust necessary for a strong customer experience.

Post-script to the Shopify Case

In the months following the initial memo, the company’s approach has shown a more nuanced implementation, signalling some adjustments to address widespread criticisms, both externally and from staff.

Focus on Human Oversight: Shopify’s “Robot & Agent Policy” was updated to require a “final human review” for any purchases made through agentic AI platforms. The policy explicitly states that “checkouts are for humans,” clarifying the importance of human oversight in transactions to maintain security and customer trust.
Enabling Internal Tools: Rather than simply mandating AI use, the company has heavily invested in providing the necessary infrastructure for employees. This includes an internal LLM proxy that gives employees easy access to various AI models from a single interface. This tactic removes friction and allows for a culture of experimentation where employees can prototype their own AI solutions.
Promoting Human-Centric Skills: The mandate has been reframed to highlight the value of uniquely human skills. Employees are expected to demonstrate skills that AI cannot replicate, such as “creativity, complex judgment, and empathy”. This positions human workers as partners to AI, focusing on the higher-order tasks that the technology can’t perform.

Universal Adoption

Yes, adoption needs to be universal; it is not just a recommendation. But the plan has to facilitate the humans in the business.

A Framework for Action: Seven Steps to a Successful C-Suite Mandate

Based on the evidence, the following seven-step framework is recommended for any C-suite leader considering a top-down AI mandate:

Articulate the Vision and Secure Buy-in: The C-suite must clearly define the “why” of the mandate, linking it to core business strategy and ensuring a fully committed board. The purpose should be to gain a competitive advantage and rewire company operations, not just to reduce costs.
Define and Empower AI Champions: Follow the Norges Bank model by identifying enthusiastic employees across departments who can provide peer-to-peer support and drive grassroots adoption. These individuals are crucial for normalising AI use and building cultural momentum.
Invest in Comprehensive Upskilling: Launch organisation-wide training and reskilling programs tailored to different roles, positioning AI as a career-critical skill. A mandate without a plan for training will be met with resistance and will likely fail.
Prototype and Start Small: Avoid a wholesale, risky implementation. Instead, begin with small-scale pilots that demonstrate clear, tangible value and allow for an iterative, “build our way forward” approach. This helps employees build confidence and provides a safe space for learning from mistakes.
Establish Formal Governance and Ethical Principles: Implement a robust governance framework with clear policies on privacy, bias, and human oversight. This is essential for building and maintaining trust with all stakeholders and ensuring regulatory compliance.
Create a Culture of Open Communication and Feedback: Acknowledge employee anxieties, celebrate successes, and create forums for open dialogue. A mandate must be a conversation, not a monologue, to gain genuine buy-in and address concerns.
Embed AI into Performance and Planning: Integrate AI proficiency into performance reviews and strategic planning, making it an explicit metric for career advancement and resource allocation. This ensures that the mandate is a long-term commitment and not just a short-term initiative.