Harnessing AI Large Language Models for Process Improvement

Artificial Intelligence (AI) is on everyone’s lips these days. In the realm of AI, Large Language Models (LLMs) are emerging as pivotal tools capable of understanding and generating human-like text. However, the capability extends far beyond mere text generation, offering businesses an opportunity to leverage them for process improvement.

Have you considered how your businesses might utilize LLMs to enhance and streamline your processes? Here are five ideas to get those creative juices flowing:

Automated Data Analysis

One of the primary uses of LLMs is analyzing textual data. Businesses can feed feedback forms, surveys, or customer interaction transcripts into these models to identify process pain points or areas that require enhancement.

Working with a Fortune 100 company, we reviewed thousands of customer comments and complaints associated with its human resource information system. We searched for and identified specific themes and then measured how often these themes occurred. This helped the systems improvement team identify and prioritize pain points to work on.

This activity occurred before AI LLMs were around, and the data analysis took days of mindlessly pouring over reams and reams of text-based comments and complaints. Today, an LLM could quickly gather the top themes and then count how often those themes occurred. This could be done in minutes and could be analyzed regularly.

Process Documentation

Incomplete or unclear process documentation can hinder operational efficiency. LLMs can assist in generating comprehensive and descriptive content, ensuring that processes are meticulously documented and easily understandable.

Although LLMs, at this time, cannot draw process maps (perhaps not far off), they can help generate very clear and specific process documentation for standard work. Also, I have found that LLMs do a good job reviewing written documents, and they could be used to review written process documentation and provide advice.

Training

For businesses aiming to ensure consistent process knowledge across their teams, LLMs can be invaluable. They can create training materials and even respond to frequently asked questions, aiding in the seamless onboarding of new employees.

One of the more significant change management efforts associated with controlling process improvement changes put in place involves process training. LLMs are very adept at developing training material quickly. When coupled with some of the AI text-to-speech capabilities that exist today, complete training videos can be built in no time and without the help of professional course designers.

Benchmarking

LLMs, like ChatGPT-4 and its Code Interpreter, can be employed to analyze and compare a company’s process data with industry benchmarks or best practices derived from textual data sources.

The American Productivity & Quality Center (APQC) provides a great deal of process data through its process framework and benchmarking activities. This data can be quickly compared to your operational data. Some other areas that are ripe for benchmarking are:

• Product Feature Analysis: Analyze product descriptions, reviews, and specifications of competitors’ products.
• Customer Service Standards: Analyze customer feedback from industry leaders.
• Content Marketing Strategies: Study successful content marketing strategies within its industry.
• Operational Procedures in Manufacturing: Analyze operational procedures (safety protocols, production techniques, or waste reduction methods) documented by industry leaders.
• Compliance and Regulatory Adherence: Industries, like finance or healthcare, can compare their policies with industry standards.

Idea Generation

Given a set of inputs or data, LLMs can generate a plethora of ideas or suggestions aimed at process improvement. Incorporating LLMs into brainstorming does not diminish the importance of human creativity and intuition. Instead, it provides a complementary tool that can amplify the ideation process, ensuring that businesses remain innovative and agile in addressing challenges and capitalizing on opportunities.

Traditional brainstorming sessions involve team members collaborating to share, discuss, and iterate on ideas. While this human-centric approach is invaluable, integrating AI LLMs can add a new dimension to this process.  They provide a diverse perspective based on a vast database of information. Instead of waiting for team members to think and articulate, LLMs can instantly generate multiple ideas based on given prompts. Humans are naturally prone to certain cognitive biases that might limit their ideation process. LLMs, being neutral, can provide suggestions without these biases, ensuring a more objective set of ideas. Additionally, LLMs can be programmed to generate ideas in a structured manner, focusing on specific areas of interest or following certain guidelines.

If the LLM is provided data, such as customer feedback, sales figures, or performance metrics, LLMs can generate ideas that directly address identifiable issues or areas of opportunity. While traditional brainstorming sessions have a start and end, LLMs can be used continuously, constantly generating ideas based on new inputs or evolving data.

A Word of Caution

While the capabilities of AI LLMs are proving to be vast, it is imperative that they should complement, not replace, human expertise. For optimal results, a symbiotic relationship between human understanding, industry-specific knowledge, and AI tools is essential.

It is clear, however, that AI LLMs can significantly help in the realm of process improvement.

---