Unlocking the Power of Unstructured Data

In the world of modern business, the ability to swiftly adapt and innovate is a necessity for survival. And nowhere is this more true than in markets of micro-, small, and medium-enterprises (MSMEs) around the world. These markets are often difficult-to-navigate, inefficient, and daunting for our entrepreneurs. Many tools like financial tracking could help achieve transparency and structure in this world, but are inaccessible or tedious for a variety of reasons. We're actively working on this problem, and you can read more at our previous post here.

Amidst this landscape, there is an amazing source of untapped potential in unstructured data. Unlike its well-organized counterpart, structured data, unstructured data is the chaotic, raw information that doesn't fit neatly into traditional databases. Think of it as the difference between a neatly filed cabinet and a room full of scattered papers, images, and recordings. Yet, within this 'room' lies insights that could revolutionize how small businesses and entrepreneurs operate and thrive. In the world of MSMEs and Enterprise Support Organizations (ESOs), unstructured data is even more prevalent, and the opportunity more vast. Despite our best efforts to carefully curate our data, it's often less structured than we wish, or miss out on the full potential of all the notes and documents strewn around our activities.

This unstructured data is everywhere, from the emails you send and receive to social media interactions and beyond. In fact, it's estimated to represent over 90% of the world's data and remains largely unexplored by small businesses. However, the advent of Large Language Models (LLMs) like GPT has begun to change the game, unlocking new ways to process this data and extract valuable insights. And this represents a key time for us to reimagine and reinvent how we think about our business support systems and impact tracking to fully leverage this new technology.

At ONOW, we are actively building tools that leverage the richness of unstructured data to deliver more personalized content, superpower our mentors and coaches, and provide a coach for every business around the world - and all at an approachable cost. If we have any hope to provide a meaningful solution at a reasonable cost, we must start to utilize the 90% of our unused data using these new tools and techniques. The following are two short examples of applications we're exploring and building. If you would like to learn more, please sign up for our latest updates here. Or send us a message and we can talk about how to achieve these outcomes for your operations and business owners.

Example 1: Personalized Learning Modules for Micro-Businesses

Historically, many ESOs carefully curate their training content with defined learning journeys and pre-determined goals to achieve. While this works for traditional schooling, it often fails to account for the complexities of a variety of business types and competency levels. Imagine a small craft business and a fishing operation —these businesses have vastly different needs and could benefit from bespoke examples and action steps.

This programming can be collected in an unstructured PDF format, given to a variety of AI-and LLM-based processes that keep the best parts of the control and structure, but dynamically inject personalization for all types of businesses and student competencies. It also allows the user to dynamically ask questions on the material and answer in a natural and engaging way.

*The Budgeting Basics course found here is an example of "unstructured data" that can be dynamically applied by LLMs in a way that would require careful structuring and curation in the past.

Example 2: Coaching Support for Business Mentors

Business mentors often rely on coaching calls to provide guidance. These calls are a goldmine of unstructured data, and nearly every ESO struggles to understand what's happening during these interactions. The notes taken during a coaching call or group educational session are a huge source of untapped insights that could be used to report on impact, improve operations, or help the mentor dynamically adjust their approach.

LLMs can transcribe, analyze, and summarize these sessions. No longer do we need to create careful rubrics or data-entry files to understand what's happening. We don't need to train mentors on data entry and work through data quality issues. Instead, we can simply direct them to enter their notes and thoughts, maybe give some general directions about areas to include, and extract actionable insights from a simple notepad or voice recording. This could include identifying common challenges, monitoring progress, and tailoring advice to specific needs, thus significantly enhancing the support mentors can offer. All of this was a pipe dream before the recent advancements of LLMs, and it unlocks a wealth of possibilities that were unimaginable just 3 years ago.

*This contrived example uses a mixture of structured and unstructured notes to create dynamic summaries in a human-readable format for a business mentor to prepare for a call with this business owner.

Let's Embrace the Unstructured Mess Together

The integration of LLMs into business operations opens up a world where every piece of data, no matter how messy, has value. Instead of being an overwhelming and dauting task to manually sort through it, LLMs can do this in an automated and dynamic way. For small businesses and entrepreneurs, especially in developing countries, this is not just an opportunity; it's a gateway to innovation and growth.

Exploring the use of LLMs to process unstructured data might seem daunting, but the potential rewards are immense. We encourage ESOs and MSMEs to explore this technology and talk with us more if you're interested in transforming your operations and unlocking new potential. Please join our community and sign-up for the latest updates here.