Divya Iyer

Although AI agent is usually an effective mechanism, being more reliable and intuitive will go a long way in adoption and consumption of such AI products. In respect to supply chain management, the area of error or delay in shipment or delivery can be explored as a scenario: 1. Response time, tone & feedback: - In case of any delay noticed in delivery, the AI should not only highlight the fact that it is delayed, but it should display empathy and provide an alternative/easier solution - This response quick in nature will develop a positive relationship between supplier and customer but at the same time it should validate the same e.g. materials shortage) - Switching between voice, text, and image will be an added advantage 2. Error- handling: - The AI should be able to pinpoint the error in the prompt from cthe ustomer rather than dismissing it stating its an error - Provide the customer with similar or closest possible responses as alternatives. 3. Proactiveness: - The AI should be able to analyze and provide suggestions or judgments in case of any cost-saving possibility between multiple vendors.

In supply chain process, AI hallucinations can lead to serious issues like incorrect inventory forecast, incorrect or misunderstood contract terms and agreements. In some extreme scenarios this might also lead to financial and reputational loss Few steps that can taken to mitigate such situation is: 1. Ensure the AI retrieves data only from verified sources of supplier database. 2. Cross verification in terms of requirement and availability of products. 3. Ensure there is transparency by mentioning when the AI generated information is sourced from predictive models and not verified sources and vice-versa. 4. Human intervention in case of any red flagged scenarios. Overall focusing on correct source, verification and strict workflow can help in evading major issues.

In supply chain management, AI-powered agents will help ensure smooth operations and mitigate compliance risks. Few areas where the AI agent can support and mitigate compliance risks: 1. Identify fraudulent payments, duplicate payments 2. It can ensure adherence to regulatory policies in terms of safety and achieving industry standards 3. Red flag any supplier basis their market stand and ability to keep the relationship intact by assessing their financial and market stand 4. Predict any disruptions In a non-intrusive way - it can ensure prioritize the necessary changes in the workflow and ensure that t the supplier follows and aligns with the agreed standard.

In supply chain management, an AI solution that indicates to the customer any failure or delay in logistics/shipment is a good use case. In case of scheduled delivery of products to the customer, if the AI can understand or predict a delay in shipping due to external environmental factors, vehicle maintenance delay based on historical records, or due to weather conditions, it can recommend : 1. an alternative route, 2. different slot for delivery, 3. suggestions on how to address the delays, 4. notify the warehouse for the backup vehicle This support from AI can help save unnecessary cost leakage due to delays and ensure near-perfect turnaround time.

1. Using conventional AI models and methods (such as rule-based systems or classical machine learning) - Conventional AI methods includes If-then, decision tree, classification etc - Easy to implement with limited database and less complex structure; - There may be a scalability issue when it comes to larger database and handling unstructured data - Suitable in scenarios like - Spam email filtering, sales prediction etc 2. Fine-tuning an existing Large Language Model (LLM) to specialize it for a domain-specific task - - Fine tuning is a best fit when it comes to working on specific task/dataset - This method is best fit instead of working on a model from a scratch - The disadvantage in using this method is requirement of high skillset and domain knowledge 3. Training a new AI model from scratch using raw data and a custom architecture - Training a new model from scratch will enable a customized approach basis the client requirement - Similar to fine tuning this method also requires high skillset Both Fine-tuning and Training new AI model is suitable in cases where high precision/accuracy is required. 4. Designing solutions with flow and prompt engineering without retraining the underlying LLM - Taking the benefit of pre trained models, this method is a very quick implementation approach and more adaptable - Chatbots for customer support or agent queries can be easily implemented using this method. Conventional AI models and Flow & prompt method is the best way for quick and easy implementation scenario whereas fine tuning and training from scratch is suitable for customized high accuracy scenarios

*Relating to the Organization that I work at - AI agent for Order processing & shipping process. * Current Scenario - In the order processing area, currently we have AI agents introduced to perform functions in silos and only in certain required areas leading to partial AI+ partial manual/RPA model. The primary reason being - multiple source tools, unstructured and unclean data management issues without a fix, incapability to address multiple, prioritized changes etc. Future Scenario - Without any technical limitations and assuming all the above-mentioned hurdles will be addresses by the advanced AI agent, my idea is to create - E2E order processing by AI agent. This AI agent should be able to cover the - Verification - Verification of Client ID, Name, Address, Payment mode, Payment details etc. Product availability - Automated check on the product availability Inventory check - Post selection of any products, if the inventory goes below the minimum level, it should be able to highlight and trigger the inventory management. Prioritization - Prioritize the request basis urgency, location, time to deliver etc. Future requirements of clients - Forecasting the customer requirement basis the historical purchase behavior pattern. Address product return issues basis pre-defined checks and audit. Suggest competitive discount prices that can be offered by the company basis industry benchmark This AI agent will interact with Humans to close the loop on feedback for improvement, address the constant gaps/escalations to improve it further. Assisting on any issue real-time if need be. Although the expectation is to enable an E2E AI agent for this process, there are certain compliance check points (e.g - GDPR) that requires constant audits, risk flagging.

*Relating to the Organization that I work at, this situation arises in case of return of defective IT spare parts delivered to the customer. * Once the order is delivered to the customer, as per company policy, there is a 30/45/60-day return/exchange period triggered depending on the product/material. For the AI (Dispute AI), any queries raised post the scheduled open window will be rejected. An AI will not be able to understand the human reason behind the delay (personal issues etc.). This approach may lead to damage to the engagement model that the company has been following over years. Earning the trust of the customer will be difficult if the company does not show an accommodative approach to the clients (especially long-term clients). Although the AI proves to be a useful tool in responding to and addressing the disputes of the customer, for such cases there needs to be a line wherein human intervention should be brought in to address such ethical dilemmas. Suggestions on the same: 1. Identify/prioritize long-term/high value customer 2. For such important customers there should be a buffer policy, or an extension period provided to ensure good customer relation/satisfaction 3. In case of such disputed scenario (similar to the example above), there needs to be an escalation matrix defined in order to have a human intervention at some point to derive a realistic solution. These checkpoints will ensure fair and equal approach for customers based on their historical order value (revenue), volume, relationship with the company and thereby ensuring a balanced outcome for the customer as well as the employee.

In an example of fulfilment Ops in my current organization, we have 3 AI products (1. Invoicing; 2. Payment support; 3. Dispute). The invoicing AI enables seamless invoice submission with the necessary, inputs from the core system like the Customer ID, product, payment terms etc. The expectation from Payment support AI is to classify payment query emails, extract information from the source systems and generate email responses. The Dispute AI looks into the query and extracts the required info and recommends solutions basis the disputes raised by customer and pulling data from source In the above case, the primary requirement is to ensure that there is a common source of information enabled for all the above 3 products without which it would be challenging for the AIs to generate the right level of information to the customer query. e.g. - If the invoice details and payment terms are not aligned the, the support AIs will pull incorrect information to respond to queries or resolve disputes. Ensuring the source is common, real-time updates enabled for the source and a standardized communication template is designed to ensure standardized response and query management.

In a Finance Operations scenario, wherein for payment of a completed order, AI has been enabled to auto populate the bank account details of the Customer for payment to be made. In case of any error OR multiple account details updated in the database, AI may consider the incorrect details and payment may be pushed to a wrong account. The financial impact caused due to this error may vary with variable impact In the above mentioned scenario, human reviewer and database creator will be assigned responsibility. One of the safeguards that can be enabled in such a situation - Strict governance around the data updates along with regular audits and continuous enhancements and updation.

In Sales Operations process - Quote Cycle time is an important parameter to ensure timely and accurate quotes for clients thereby ensuring good customer satisfaction. In this scenario, AI can help in ensuring correct inputs and mapping (e.g. Product #, Client ID, Quantity etc) are logged into the quotation basis client requirement. By ensuring the right level of inputs, it checks the quality of the inputs and timely submission of quotes with minimum turnaround time.