Ankur Singh

I can very well relate to an active problem statement I am working on - Disabilty Claims allocation to Examiners. LLS approach to improvise allocation of claims : Process was as below- Indexing > Claim Set-up > Determination of O/S requirements > Claim Examiner review > Ask/follow-up on O/S requireemnts > Claim Adjudication > Payment I was realized that there were several hand-offs in the process, lack of expertise of resources at different stages - Indexer, Set-Up associate and Examiner. This lead to lack of determination of requirements early on in the process. The process was improved to reduced cycle time and improved process SOPs/documentation at early stages of the process. Reimagination : We are aiming to replace manual triage of dcuments to an OCR/NLP based Input layer at indexing stage, which will 1. auto index the claims. 2. determine all outstanding requirements via a Rule Engine (AI enabled) 3. Automatic Claim Set-up and follow-up emails to claimants for O/S requirements 4. Assigning claims to Examiners basis their Skill-set, Territory, Capacity, avalabilty, etc. (AI enabled, along with integration of ERP system) 5. STP for simpler claims Eg. Materninty. Additionally, AI can also help in generation patterens in data like- Rejection rate, Over/Under payouts, Escalations, Follow-ups needed per claim, etc.

1. Like any project first step is to create and Impact - Effort matrix to prioritise solutions. 2. Do a PDCA for quick small scale implementation and pivot the solution basis observations. If possible an A/B testing can be beneficial as well, as learned during CAIPO 3. Managing change w/o resistance is a critical, and will require clear communications, leaders championship for changes, involving high influencers in team to be looped in into the plan early on, etc.

Like any other project- during Analyze phase, we must first define the problem well, and thereafter break the process into steps to determine bottlenecks and root causes. Approach- 1. Prepare the current state process map. 2. Do a few brainstorming sessions to identify potential causes.(Ishikawa diagram and 5Whys) 3. Map the potential causes to steps/stages in the process map 4. For shortlisted potential causes do few hypothesis testing to validate the data/causes. This approach will help us separate real causes from noise.

For this situation, In measure phase we can do below - 1. Focus on metrics which are directly associated with CX, eg lead time, wait time, 2. Avoid putting too much efforts on metrics which do not have significant relationship to process, eg- # of patients in a day 3. Use a combination of metrics to identify relationships. Eg- paitient wait time along with Patients VoC and Diagnosis. To catch bad/incorrect data- - The data entry tools for staff can be forced to accepted data basis historical experience/trends. Any outliers must be highlighted by system at time of entry an requires a forced entry, if needed. - Such anomaly of data must add a flag in the system for subsequent stages of patients journey for review and validation.

Leveraging VoE for Business Excellence - In may experience employees of large organisation get too many surveys that are VOEs. For BE, it must be designed in a way taht employees are encouraged to share Continuous Improvement ideas, Business Pain points, Employee pain points, Bottlenecks or inefficiencies, etc. A reward can be helpful for most contributions from individuals. Challenges - Survey insights are mostly feelings/sentiment based. Data to back these statements are lacking. Respnse Rates are poor, and often there is too much Noise in data. Overcoming Challenges - To improvise VoE we must ensure leaders champion the initiative, offer support to employees, protect anonymity, Share regular updates with employess on their feedback, Use data/text mining to identify trends. Most importnatly Recognize who contribute regularly and signifincantly to the excercise.

I believe, there is no definitive answer to this question. Both the approaches are fundamentally different, and in several circumstances both can me used togeather. Specific needs and capabilities to determine the most appropriate methodology or combination of both must be assesed by organizations. BPR: Usually involves a shorter, intense period of change with quick and more disruptive results. Lean Six Sigma: Requires a longer-term commitment to ongoing improvement, with results improving over time. Both can be used combination in a situation, wherein, The BPR can be leveraged to brign disruptive cahnges in the process, and thereafter LSS must to applied to sustaion and incrementally improve the newly designed processes. I major differentiation can be the cost - BPR as ususally cost intensive, while LSS can be deployed with a a comparativly less cost.

As an RPA consultant for my organization, I would include following UX design choices on AI agents; - It should use user's name and remember previous conversation history to make it feel personal. - Another customization that can be great is asking for a language preference from the user. - It should offer next questions for each response basis the agents memory to offer help proactively - Agent can be tasked to determine sentiment of the user during the session, and responds sensitively - Graphic/Visual clues for difficult queries could be helpful. - When the AIagen hands over the conversation to a Human Agent, it should summarise the conversation for Human Agent to avoid repetition of information capture - Lastly, agent must be tasked to avoid being 'over cheerful', this is one thing a observe with most of the genAI chat bots. It feels robotic.

Being associated with an Insurance Company, below are few things we can do to identify hidden patterns- -Build a repository of all issues, tickets, escalations, errors, etc. -The repository must be structured to ensure capture Department, Issue Category, Time/Date, Process/Function, Product, brief issue, etc. -Place an AI agent to collate all the issues at one place, via connecting through different data sources -Run correlation tests via AI agents which are tasked to identify patterns. This will require giving specific persona and prompt while creating the agent. -Schedule the agent to initiate an email with this analysis and visual representation of these patterns.

North America Group Health Claim Processing - Requires 3 documents, Employee claim Statement, Employer claim Statement & Attending Physician diagnosis Statement (APS) Exception Scenario - Missing documents/info Description - In as claim operations office claims are generated with either of these docs received by Ins. company. However, SOPs advise to gather all three documents before processing the claim. In several scenarios, Claim Examiners take/provide exception approval, to proceed with claims in lack of all the information. For ex – APS can be skipped for Maternity claims, etc. Why - AI can be leveraged to filter claims for Examiners to proceed or send back to claimants, basis previous decision patterns. How - After every claim decision, force examiners to input/tract below details. What - Data to track > Missing Info: Y/N, Type of Info missing: <text>, Decision: <Proceed/Send Back>, Rational for Decision: <text>, etc

Save the Driver or Save the Pedestrian -Scenarios which are Critical/Security related where the decision can be ambiguous. For ex - If a man suddenly comes in front of driverless driven car, who should AI save the passenger in car or the pedestrian. In similar situation replaces the man with a 5 yr old child. WHAT should AI do in both situations?