#05 RAG: Building a Durable Competitive Advantage with Your Expert Insights

Welcome back, dear reader! The corporate beat is less like music and more like a smack to the backside, so I hope your Monday has been gentle on you. Now, let's escape into some reading!

The concept of leveraging AI technology to enhance efficiency and creativity spans various levels, from personal career advancement to enterprise innovation. Let me illustrate how different environments can benefit:

• Entrepreneurial Level: For entrepreneurs turning innovative ideas into scalable businesses or keeping existing ones relevant, rapid prototyping and quick feedback loops are crucial. Access to comprehensive market insights and predictive analytics can empower data-driven, proactive decisions that differentiate you from competitors.
• Company Level: Companies operate in fiercely competitive markets with a shareholder-driven approach. AI and robotics provide tools to create more value with fewer resources. However, this often leads to unprecedented job destruction, with some new roles also at risk of eventual AI replacement. Tech giants are already downsizing as they lead AI implementation.
• Employee Level: Employees face competitive workplaces where innovating and accessing timely knowledge can set them apart. Tools enhancing personal learning and decision-making capabilities help individuals exceed role expectations.

?? Got 15 minutes? For those who have been following along from the beginning, I am not the only one saying that for employees, these corporate benefits are typically not good news in the long run. Watch this video on YouTube to know more (John Stewart, The Daily show, April 2th '24).

• The "Autonomous Enterprise" Level: Ultimately, the only thing setting us apart from AI may be knowledge we haven't explicitly provided it, and novel ideas based on that knowledge. This is why I previously advocated for the Autonomous Enterprise - letting you focus on great ideas while AI handles operations. Knowledge management and governance are key, as they typically distract from your core business.

I believe as employment becomes harder, entrepreneurship should get easier. "Entrepreneurship" simply means undertaking something meaningful driven by your passion. Then we can tap into the collective intelligence of billions once we enable turning ideas into reality. The full story you can find in my previous blog post:

But in any situation you are in, Retrieval Augmented Generation (RAG) emerges as a crucial "moat" - what venture capitalists call your competitive advantage that can't be easily replicated. The knowledge you've built, unique opinions that led to your success, proprietary datasets - these things set you apart, and understanding RAG is vital.

What is RAG? It's providing your data or knowledge to an AI system, leveraging the AI's reasoning capabilities rather than just its pre-loaded knowledge. Why RAG? Most public knowledge is now ingested by large AI systems, including potentially outdated or misaligned information. Your intelligence and proprietary insights should take priority as key differentiators.

RAG unlocks AI's reasoning talents on your data, so let's quickly repeat what kind of talents emerged once language models got really large:

• Transformation: Reformatting, refactoring, language changes, etc. Input ≈ Output size.
• Expansion: Drafting, brainstorming, amplification from an initial idea. Input < Output size.
• Reduction: Summarizing, analyzing, evaluating larger inputs into concise outputs. Input > Output size.

Large Language Models (LLMs) have no explicit knowledge or hand-coded rules, all the above reasoning talents are emergent results of recognizing patterns in their training data. This amazing article by David Shapiro contains further details.

I'll provide an example from personal experience:

Say you have a strategic plan as a PDF. An AI can critique it for consistency, ensure all parts align over time (it understands temporal concepts), check you're not drifting off-topic, etc... All of this feedback combined can feel like a "death by a thousand cuts", but once addressed, you have a robust, management-ready plan.

Different roles and sectors can leverage RAG in unique ways. I'll focus on the main industries my newsletter followers are in:

• Technology & Innovation: Cybersecurity, innovative IT problem-solving, AI-assisted development, evolving user-centric applications.
• Business, Finance & Marketing: Strategic planning, comprehensive data analysis, personalized marketing strategies.
• Public Sector & Education: Evidence-based policymaking, effective public communication, adapting education to future needs.

For now, consider: What would you ask your data? What problem centrally involves your expertise? What entrepreneurial passion could your access to information and AI's reasoning talents enable? Your unique knowledge will be a key advantage as large language models commoditize general information.

In my previous newsletter, I mentioned that apart from tapping into your own knowledge through RAG, agents are another key component to consider as they ensure that tasks can be executed autonomously. So expect a post like this one, but focused on agents.

Next article:

The manifesto: