Rupinder Narang

Prompt and Flow offers the option to design (typically using LLM prompts), test, fine tune and finally deploy your AI agent, allowing you to follow an iterative process and make improvements until the final outcome meets expectations. To spot common patterns across escalations, issues, complaints across various functions, a common database to log all the comments/verbatim can be created. A prompt can then be generated to search the internal database (knowledge base) and return answers such as- - what is the general sentiment in the comments (refer to the column header with verbatim or commentary) - are there common themes of comments such as delay in resolution - Is there any mention of behavior issues in the comments - how do comments for different departments compare against each other - are there more complaints on any day of the week etc In order to make it easy for the leaders to act on, the output can include summaries like count of negative themes by department and sample comments. The most critical input will be a well structured database using which the prompt + flow model can fetch answers. Some of the fields in the database can include the date of issue raised, name of the department, commentary, etc.

Amongst the things that AI cannot do today, thinking, feeling, emoting, answering questions that involves these faculties are on top of the list. AI also cannot generate AI. It cannot answer hypothetical questions either. If I would imagine what I’d like to be different, that would be AI generating AI - something akin to problem solving but without training data, without any AI training. I’d like AI to be able to spot problems and generate solutions autonomously- examples how to be energy sufficient across the world, how can everyone have enough water, how can the richer countries be prosperous without exploiting resources from other countries, what are the diseases that may emerge that haven’t been given enough attention. The gap could be that even if AI started doing all of this - it would need to do more - such as be able to detect bias in data, be culturally sensitive in proposing solutions, be able to remove actual biases such as lesser data availability for under privileged races or countries. Failing this, all the solutions that AI generates would be ridden with fallacies.

There are multiple examples of multi-step AI or reasoning agents - such agents break complicated tasks into smaller, manageable chunks which are carried out in a logical sequence. The tasks must happen in a sequence - example - you cannot start cooking until you have the right ingredients. And while you can make adjustments while cooking if you realize that you are missing an ingredient, AI needs to have the path clearly outlined to complete the process end to end. To mention a few examples: 1. AutoQA in Call Centers: The first AI agent will convert voice into transcription using ASR, the second agent will analyze the data collected to identify key quality elements using LLM, yet another agent will utilize the identified quality elements for scoring against a yardstick, and there could be another that analyzes deviation from expected norms 2. Self Driving Cars: The car sends out LIDAR signals and the data captured from its environment is compared against the mapped data, this acts as an input for the next step in decision making of the action the car should take e.g continue driving, slowing down or coming to a complete halt, plotting a map of the trajectory for the car to follow, and sending that information back to the cars actuators Such agents operate using 3 different methods: Chain of Thought (COT) - Problems are broken into steps, with calculations for each part of the sequence outlined ReAct (Reasoning + Acting) - Here the AI agent uses reasoning to take real time actions, allowing complete interaction with environment and making adjustments to the process based on feedback. Each reasoning step is followed by an action step Reflexion - Feedback loop iterations are used to enhance the model by using past output as learning inputs Some of the challenges could be resource usage, unclear rules for coordination and conflict resolution. In order for a multi step agent to work in a coordinated manner, the goals must be clear, agent design is usually modular, communication protocols must be well defined. Accuracy and speed of each modular element must be assessed in order for the logical sequence to work well. Communication protocols must outline how the agent shares information, coordinates actions and resolves conflicts. There must be governance system for identifying and resolving issues within the system. One of the other ways of ensuring coordination is making the flow sequential to avoid conflict.

This question challenged us to go beyond what we have frequently heard in most Lean and Six Sigma discussions, which is that Rework should be avoided or reduced or eliminated, where possible. The contributions show how this group is open to evolving ideas. "Nothing is absolute. Everything changes, everything moves, everything revolves, everything flies and goes away” - Frida Kahlo. The first part of this statement is applicable to all subjects, including, “Rework". Very creatively mentioned by Mohan PB, who has also supported his answers with diverse examples. This is the chosen best answer for this Monday's question. Other answers also cite some examples, which are worth a read.