Generative AI is everywhere. It dominates headlines, conferences, and debates, while fueling both excitement and concern. It can already automate certain tasks, accelerate data analysis, generate content, and transform jobs and business models. Yet many Quebec companies are still hesitant to move forward, not because they lack interest, but because they are unsure where to start.
For us, the main challenge is not the technology. It is the strategy. In an economic environment marked by volatility and uncertainty, digitizing and automating more is not enough. Companies first need a robust business model and processes that can perform well under normal conditions while remaining resilient when disruptions occur.
Before choosing an AI tool, we therefore need to determine where AI can actually create value. A good place to start is by mapping the value stream, from customer demand to payment, using Value Stream Mapping (VSM). This systemic view helps identify wasted time, quality issues, bottlenecks, vulnerabilities, and areas where AI can make a meaningful contribution.
Automating repetitive tasks, generating content, summarizing data, or simplifying follow-ups can all create value. But automating a useless activity only produces waste faster. AI integration should therefore build on the fundamental principles of operational excellence and robustness.
We use five principles to design processes that are both high-performing and robust:
1. Clarify what truly creates value for customers. Anything that does not create value should first be challenged or eliminated rather than automated.
2. Standardize, control, and continuously improve critical activities, especially bottlenecks and activities that are essential to quality, while preserving greater flexibility elsewhere. Performance depends on discipline where it matters most, while robustness depends on the ability to adapt when conditions change.
3. Avoid maximizing the utilization of every resource. A company with no spare capacity quickly becomes vulnerable to even minor disruptions. Maintaining buffers, particularly upstream of bottlenecks, makes it possible to absorb variability without destabilizing the entire flow.
4. Implement real-time visual management of the value stream. The sooner a performance gap or vulnerability is detected, the more options the company has to respond before the issue spreads.
5. Support visual management with a continuous improvement process. Employees need to be trained, equipped, and empowered to understand gaps, test solutions, and improve their processes.
AI can strengthen each of these principles. It can accelerate analysis, automate certain activities, and make information more accessible, but it does not replace strategy, human judgment, or collective intelligence.
The question is therefore not, “Which AI tool should we use?” but rather, “Where can AI genuinely improve our value stream without adding unnecessary complexity?”
