Airat

In a process of Supporting inventory visibility and stock accuracy in a manufacturing environment. Using the Swiss Cheese model, I can see that no single control is perfect because risks can still pass through weaknesses in multiple line ups. Below are examples of Defense Layers in this processes: Real-time stock tracking system Physical stock counts by warehouse staffs Approval workflow for material requests before the start of a process and during the process. Audit checking by the inventory control team of the company. Weaknesses(Holes in the Cheese): Data entry errors or delays in updating the digital log Incomplete or rushed physical counts during busy periods Bypassing approval dur to urgency or familarity Audits being skipped due to manpower shortages. Below decribes how this guides Business Excellence Improvements: Using the Swiss Cheese Model, I can apply Business Excellence principles like the Poka-Yoke (error-proofing), Standard work, and Root Cause Analysis to plug those holes. For example: the Automation of reminders for stock count deadlines Rotation of Audit responsibility among staffs to ensure it gets done consistently the Introduction of barcode scanning to reduce entry errors. This thinking will shift the mindset from fixing the one big issue to Strengthening the system as a whole.

Change management is really relevant across all phases of DMAIC, this is because no matter how good the solutions and data are, people are still the major factors that drives change. Below are examples of management at each DMAIC phase: Define Phase: In this phase, the MBB should ensure stakeholders understand the WHY behind the projects. Poor alignment in this phase will cause resistance later. For example, the reduction of customer complaints in a Nigerian manufacturing firm. Early alignment with the sales and quality team must be done to make it easy to collect data and test ideas without friction. Measure Phase: In this phase, people may question the data or feel defensive. The role of the MBB should be to create transparency by involving frontline staff in validating the data and ensuring trust. For example; consider a delivery staff who was initially resistant to GPS tracking during a logistic project. those delivery staffs can be brought to co-review the data and understand that GPS trackers isn't for blames but for route optimization. Analyze Phase: here, MBBs must frame insights in a non-threatening way, showing how fixing issues helps people do their jobs better, not replaces them. For example, Imagine a delay was tracked to manual approval. MBB can work closely with managers to see automation as support and not loss of control. Improve phase: The MBB plays a vital role in piloting changes, gathering feedback, and making people feel heard. Control Phase: Changes fades without reinforcement, so the MBBs should ensure standardization, ownership, and monitoring. Also a setup recognition reward system can be adopted to encourage continuous improvement.

Wins slip away during the control phase when teams go back to using the old processes due to lack of monitoring, accountability, or training of team members after deployment. using the logistic business as an illustration. Even if the AI solution provides a new routing system, drivers may ignore it, supervisors may stop checking the reports, or even little issues like fuel scarcity may create situations that derails the new system. To ensure improvement is in place, this tools can be integrated: Control charts: keep a control chart to track key metrics, and quickly detect any variation or slippage. and ensure the chart displays realtime performance so teams can stay aware and accountable. Process Standardization: Through simple to implement SOPs and checklists that are easy to follow and reflect on improvement systems and processes. Ownership and Handover plans: To ensure that somone is responsible for sustaining the changes on the system. Regular Audits: audits should be done regularly to ensure that what was improved is actually being followed.

To ensure that AI solutions keeps evolving progressively without misalignment, AI should be embeeded within the existing improvement cycles as a participant in the loop and not just as a seperate tool. This is very relevant in Nigeria as it is a resource conscious settings where very lean adoptions of Ai solutions must deliver a measurable value without it requiring any constant expensive retrainings. The practical integration of AI into impovement cycles: To reduce thw cost and complexity of AI soltions, Organizations in Nigeria can use low-code AI platforms that allow team leads make frequent updates to logics or just retrain model themselves without an heavy dependency on data scientists. Clearing feedback mechanism for AI adaption is crucial in Niagerian industires like manufacturing or financial institutions where policy shifts can drastically change baseline condition. AI systems are not fight and forget like the Nigeria road(haha) after a rainy season. they always require regular check ups and maintenance.

An example of a situation where AI solution might deliver technically but miss the mark is; In a FMCG company situated in Nigeria, an AI-solution can be built to predict stockouts in a retail store. assume the model gave an accurate prediction technically(that is, it could tell which locations were likely to run out of stocks) but in reality, product availability didn't improve. This is because the real problem wasn't a lack of stock at the warehouse; it was the lack of good coordination between sales representatives, transporters, and distributors. MBBs can contribute in this case, using tools like 5 whys or a process map, the MBBs will uncover issues like: Sales representative not updating orders at the right time Drivers not getting to some outlets because of bad roads or a spike in fuel prices. There's no visibility in real-time delivery. So instead of building a model that predicts stockouts, the problem will be reframed as below: "How can coordination between Sales representatives, distributions, and logistics"

To ensure that AI in a procress-driven organization is responsible and of high impact, a new AI governance framework should be included. Below are some possible new governance structures that can be included. 1. Bias and Fairness alert - There should be a constant check to detect bias in models, AI agents, and so on. 2. Post-deployment reviews: After AI models have been created and deployed, feedback should always be collected to improve AI systems to solutions continuously. 3. Continuous Monitoring: There should be an accuracy performance tracker which will track the performance accuracy and behavior over time, and whenever the model accuracy gets below the threshold, an alert should be sent. Below is the list of people who should be involved 1. AI Architects 2. Data Scientists 3. IT compliance team Below is the list of Mechanisms to maintain Agility and Control 1. Ensure that there's a traceable and rollback options for every changes or updates that's made to a model. 2. ALlow the monitoring of key KPIs such as model Accuracy, adoption rate and so on.

Below are two traditional metrics that would lose relevance in an AI-based setup 1. Processing Time for Manual tasks: As automation is reducing the time spend on some tasks and also increasing the accuracy of tasks. processes like entry of data, simple decision-making, simple critical thinking, and so on. 2. Manwork time saving: In an AI-based setup, manwork hours can be misleading, as the value of work done is not tied only to labor reduction. Therefore, AI has shifted the role a and added more value in such a way doesn't allow for manual work hours of work. Below are two new metrics that can be introduced to track real performance 1. Rate of AI Adoption in the organization: This metric will measure how widely the AI tools are used within organizations. 2. Accuracy of AI responses: This will measure how often the AI generates the responses that users want to see.

MBB should act as an intermediary between AI architecture and the Business. MBB should ensure that all AI solution is grounded in customer value and a measurable outcome for the business. MBB's role is to bring clarity on the existing processes, stating why the process exist, where the roadblocks are and how the success should be measured. MBB, should first clear process baselines before AI is developed. ensure that the AI truly brings value ensure feedback is collected after AI solutions have been deployed.

One AI solution I worked on was a conversational assistant for factory inventory and production teams. It handled tasks like checking stock levels, finding material codes, and alerting low stock items. Because it was used in a high-volume, time-sensitive environment, the UX had to be smooth, fast, and frustration-free. For tone, we made it sound like a helpful colleague — polite, concise, and never too technical unless needed. For example, instead of saying “Invalid input”, it would say “Hmm, I couldn’t find that material. Want to check the name or try the code instead?” This keeps the tone friendly while guiding the user forward. Response time was critical. We optimized all flows to respond in under 2 seconds. Where delays were unavoidable (like external sheet lookups), we used typing indicators or short status updates like “Looking that up for you...” to manage expectations. Error handling was treated as a key part of the design. Instead of dead ends, we used “soft fallbacks” that offered smart suggestions — for example, “I didn’t find that material, but here are 3 close matches — is one of these what you meant?” I also insisted on personalization. The assistant remembered users' previous queries and preferences (e.g., preferred units, frequently checked materials) to save time and reduce friction. Even small touches like greeting users by name or showing their last request made the experience feel more human. One feature I’d include in any AI solution is built-in, easy-to-use feedback. After every few interactions, a simple question like “Was this helpful?” with quick buttons gives real-time insight into user satisfaction — and it's easy to implement. In high-volume settings, speed, clarity, and emotional intelligence are just as important as functionality. If users feel like the AI “gets them” and doesn’t waste their time, that’s when it becomes a real asset.

In my area of work, and that is manufacturing support, if an AI agent provides the wrong material code or material status, production delay or wrong items are used. To reduce hallucination: I make highly specific prompts that go something like: “You may only reply with the code of what material is listed in the database.” I hook the AI up to a certified Google Sheet so it only scrapes live figures. I put in debouncer checks, like “Is it the right code? Yes/No” to allow users to confirm before proceeding. I have the logic to display the source data as well, so users can verify the info themselves.