KarthikeyanM.R.

Change Management across phases Define Understand and articulate the voice of the customer or quality, focus on the pain points or new opportunities. Building consensus on the need for change and align with stakeholders’ expectation Measure Ensuring shared ownership in collecting the data. Involve the frontline staff to baseline the data and conduct performance reviews. Analyze Collaboration across stakeholders to find the actual issue, engaging cross functional teams and involving them in detailed root cause analysis. Ensuring the transparency in publishing the detailed findings and explaining the rational for criteria being used Improve Prepare and train the team for changes Set up a detailed feedback collection process and define the procedures to assess the effectiveness of the change Control Maintaining well defined Standard operating procedures with latest updates and properly reviewed (SOP) Adopt dynamic process monitoring tools and Kanban visual dashboards. MBB Responsibilities across Phase: Articulating the problem or opportunity and providing data backed support to secure management approval. Supporting the team to establish data baseline and help to visualize current performance Guiding the project team to conduct a structured root cause analysis and supporting them to prioritize Leading the implementation and piloting of proposed changes or solution Overseeing the handover process and supporting audits and setting up other required processes for monitoring the status

Key reasons for control phase are not effective: Lack of process adherence, monitoring mechanism or real time metrics dashboards which will lead to process drifting over the time which ends up on performance degradation. Human factors, new members may not be trained to the extent to which they need to perform, so when the old guard moves away from the workforce, things will gradually fall away. Sustaining the improvement is teamwork, so without proper RACI definition in the complex process, improvements will drift away due to lack of ownership. What is required to effective control phase. Maintaining well defined Standard operating procedures (SOP) Plan for effective training for the team at regular intervals Adopting the dynamic process monitoring tools and Kanban visual dashboards. Adopting a work culture and shared responsibility which ensures commitment and focus on continuous improvement.

Define the Standard operating procedures for AI models or solution for undergoing review regularly. SOP should focus on data and process alignments along with regular human oversight (ex : Audit) Developing automated monitoring systems to assess the data or concept drifts or any performance impacts against the baseline. Implement a process to regularly collect feedback from the customer to access the AI performance. The Role of MBBs (Master Black Belts): They should act as a process champion for spearheading the continuous improvement activities for AI solution Supporting in converting insights into business requirements and leading change management for AI solution

For easy understanding let’s take one example of common development projects, Project has the following challenges, Failed to meet or continuous slippage in the delivery time timeline. Customer faces quality issues in the delivery of medium and complex requirements. Just looking at surface-level symptoms, projects teams will always tend to strengthen QA teams and decide to employ automation approaches. To fast track this automation initiative, they will tend to buy automation tools with AI augmentation capabilities or add more QA experts to support the current crisis. After deploying the new approach, teams ended up with more internal quality issues in the developments, there will be again delivery timeline delays, and the customer is going to be unhappy as there are a lot of backlogs. Let’s assume that actual root cause is compromised engineering timeline or faulty estimation approaches or critical resources shared across multiple sprints to support less experienced team. So, without understanding the actual root cause, just by looking into surface level symptoms it always leads to failure in complex or scaled projects. The role of Process excellence owner (MBB) can help to go deep and identify the actual bottlenecks to deliver the required business value.

Essentials of AI Governance: Define Key performance indicators for measuring the business value in AI driven projects. Define the process to categorize the AI project using following standard approaches Business value vs feasibility matrix, Cost benefit analysis, Risk Priority number and ensure strict compliance for the project with the highest RPN. Documentation of business logic for transparency when the stakes are high. Data compliance with international regulations (example: GDPR) Audit procedures to monitor any bias both at the training data and algorithms. Define a clear escalation process when human involvement or support is required. Define the procedures to monitor data drifts by statistical methods to compare with training data and live data or implement monitoring tools. Define different control procedures to be adopted in the case of data drift. Define continuous improvement procedures to fine tune the model based on customer feedback. Team required: AI solution architects Business excellence champions IT team Business SMEs AI Project leaders Customer Steps required for agility and control: Well defined Governance model Real time monitoring of the process AI solution change control board

Following metrics will become irrelevant when AI enabled solutions comes into practice: • Productivity measurements involving human efforts alone • Metrics involved in measuring Manual process effectiveness Metrics recommendations relevant for AI solutions • Percentage of AI Augmentation involved • Measuring successful first time responses • Percentage of human intervention

Domain: Air Cargo Logistics Process: Door to Door delivery. Sub process: Picking the material from customer > Review and complete the documentation > Delivering to nearby warehouse. Traditionally lean and six sigma is used for identifying delays, reduce wait time, optimizing the needs for additional vehicles, routes and identifying bottleneck during peak demand times. Currently most of the times it is reactive problem solving. Example : There will be delays in pick point where additional documents needed for transporting exceptional medical supplies need to be cleared. Even though clear process guidelines are shared still 10 to 15 % mismatch occurs. Here more than improvement we can reimagine by deploying AI Powered Agent which can support customer on getting the documents filled by getting required data as attachments. Sample Solution could be, customer can provide(drag & drop) material documentations available with him ,AI powered agent can fill the required forms and documents required for custom clearance and request for additional details or support for missing information’s and share the final document for review .

Recommendations for finding the actual root cause: Perform root cause analysis with 5 whys Plot the full process with swim lanes for identifying the bottlenecks and inefficiencies Interview the people Look out for patterns in the data Recommendations for avoid chasing the wrong cause: Cross check the data collected Validate the findings from different angles Discuss with people part of the process to correlate their observations

Way to decide on what metrics to considered for Measurement: Measure the complete process starting form patient check-in to discharge from hospital. CSAT survey results or ratings based on patient experience Define a target time and measure the average wait time of patients. This will help to re baseline and plan improvements to reduce the wait time . Recommendations to catch the bad data : Any major fluctuations (drop or raise) in wait time. Compare the data with other internal sources to cross verify the individual data collected. Conduct pilot runs and analyze the data ,before rolling out complete deployment in one go.

AI Enforcement: Areas of Trust (Rule Where I Would Trust AI to Make the Call) Automated Workflows: Repetitive tasks with defined urgency/priority. Consistent & Data-Driven Tasks: Objective outcomes and clear parameters. Pattern Recognition (NLP): Identifying trends and categorizing information. AI Enforcement: Areas Requiring Human Judgment (Rule Where I Would Never Trust AI to Make the Call) Contextual, Emotional, & Ethical Decisions: Nuance and human values are critical. Complex Technical Tasks: Deep understanding, hypothesis, and experimentation needed. Ambiguous/Incomplete Information: Requires interpretation and clarification. Non-Obvious Causes & Intuition: Leveraging experience and subtle cues.