2024 : Year Of The RAG
If 2023 was all about foundational LLMs like ChatGPT and Llama-2, my prediction is that 2024 will be all about Retrieval Augmented Generation — RAG.
In this blog post, I make my case for why RAG will be skyrocketing in 2024 with not just business adoption, but also skyrocketing consumer adoption.
So without further ado, let’s get started.
What is RAG?
Here is a simple definition of RAG, thanks to our friends at IBM.
RAG is an AI framework for retrieving facts from an external knowledge base to ground large language models (LLMs) on the most accurate, up-to-date information and to give users insight into LLMs’ generative process.
More recently, RAG is a way to get responses from LLMs like ChatGPT while supplying your own data. This might be little knowledge snippets relevant to your prompt, or might be user specific data from databases or transactions.
Why Will RAG Skyrocket?
At this point, the development around LLMs like ChatGPT has reached extremely mature and beneficial levels. It almost seems like LLMs have crossed over the threshold into a post-BC (Before ChatGPT) era. Some people are calling it the new “Generative Era” — aka: GE.
Today’s company is tomorrow’s product and next week’s feature. — Cesar Brea
What this means is that LLMs are now mature enough to be combined with business-specific data like knowledge bases and databases and use cases emerging from this.
RAG has far, far higher implications to businesses and consumers than foundational LLMs. It’s almost the equivalent of turning oil into the gas that powers cars.
Yes, oil is cool, but refined oil that can power a car is much cooler and beneficial in transporting people.
Similarly, when the power of the LLM is combined with knowledge, the true benefits to businesses start getting unleashed.
And that is when businesses and consumers start waking up to its effects.
From customer support, to employee productivity, to AI-enhanced workflows, the power of LLMs combined with knowledge (aka: RAG) will create tremendous revenue and productivity gains for businesses (and consumers alike!)
Who Will Be The Winners?
There are basically three categories of RAG that will be clear winners in 2024.
Category 1: No Code Systems
No code systems like ChatGPT GPTs for consumers and individuals. And business-oriented “Custom GPTs” will be clear winners as the demand for RAG-based business use cases skyrockets.
These no code systems allow everyday non-technical people to build sophisticated and complex generative AI functionality with just a browser and no coding required.
The friction and barrier to entry is virtually zero with even non-technical people being able to create sophisticated generative AI chatbots (See Case Studies)
Category 2: RAG APIs
With the release of OpenAI’s new Assistants API, which has some very limited built-in RAG and other more sophisticated RAG APIs like the CustomGPT API, businesses — with very little effort — can create sophisticated generative AI chatbot functionality and workflows using their own data, website content and account specific data.
These types of projects used to be complicated, multi month, multi-million dollar projects involving large software development teams. But now you can create a sophisticated RAG-based chatbot in less than a day and less than $100 using an Upwork freelancer.
Trust me, I have hired Upwork freelancers who have built sophisticated workflows in less than a day for $100.
There are even some Streamlit apps that I was able to create at very low cost with a quick turnaround time. As more developers start understanding the power these APIs, more RAG-based systems and workflows are going to start appearing.
If 2023 was the year of the OpenAI wrapper applications, 2024 will be the year of the RAG wrapper applications.
They might have sophisticated names, like “Custom GPTs” or “Augmented GPTs” or maybe some thought leader or journalist might even come up with a better name.
Category 3: Workflows
Towards the end of 2023, I am seeing cloud platforms like Salesforce and Zoho, all incorporating API-based workflows into their systems.
With these workflows, It becomes much easier to tap into account-level data and have RAG based workflows.
This could be something as simple as capturing an HTML form input and generating a PDF document based on that input.
Just think about dynamically generating a travel itinerary or an invoice PDF that would require some sort of generative AI component.
But PDF generation is just one of the elements. Just think of any sort of workflow wherein basic flow of data is now being augmented with generative AI content.
The enthusiasm for Large Language Models (LLMs) in 2023 was immense, but the practical applications and benefits for end users are expected to increase exponentially in 2024.
Conclusion
Remember: Joe Blow on Main Street didn’t care much about ChatGPT or LLMs, but when these RAG-based applications start reaching him in 2024, it becomes real.
Again, using my oil analogy, nobody cares about oil. They care about their car moving from point A to point B.
That is when things start getting exciting.
