There is no greater current example of a disruptive technology than artificial intelligence (AI). Its explosive adoption rates - with applications across all industries and early users openly exploring how to use it - have proven the importance and usefulness of this technology. The increasing need for businesses to automate, analyze massive amounts of data, and drive strategic actions are propelling its usage. That can be proven by the global AI tech market being valued, per Zion Market Research, at nearly $59.67 billion in 2021 and estimated to expand to $422.37 billion by 2028, which is a compound annual growth rate (CAGR) of 39.4%.
Since the launch of ChatGPT in November 2022, there has been a massive wave of active interest and usage in a specific area of AI, called generative AI, which is poised to unleash the next wave of productivity for its users, per a McKinsey recent report. In particular, it reviews benefits for Marketing and Sales areas boosting personalization, content creation, and sales productivity.
To better understand how generative AI is being applied to financial services, we reached out to Institute Founding Member Nathan Stevenson, CEO of ForwardLane – a financial technology firm founded specifically to help financial institutions leverage the power of AI. Their recently launched generative AI decision and intelligence platform called EMERGE empowers financial organizations to rapidly scale their data and analytics capabilities by allowing key sales, marketing, business, and product professionals - with no AI/LLM expertise – to easily create, preview, and engage with valuable insights. As a long-standing innovator in the AI community, we asked Nathan questions to learn from his unique perspective and experiences and take us on a deeper dive to understand what generative AI is and learn about the strategic advantages and unique capabilities it brings to the financial services industry.
Hortz: What was your motivation in launching your new EMERGE generative AI platform? How did it evolve from your core AI Insights & Next-Best-Action engine?
Stevenson: With EMERGE, we saw an opportunity to transform how businesses make decisions. For too long, critical data and insights have been trapped in impenetrable databases that only data scientists can navigate. We envisioned a future where anyone could access the power of data with the same simplicity as using Figma or Canva.
Our breakthrough came with the development of our Insight Creator tool. With conversational AI technology, it enables users to query data with a few visual clicks, and then go onto generate insights in plain, understandable language. Questions you would ask your colleague over coffee can now be posed directly to the data, unlocking a world of possibilities. But we did not stop there.
We doubled down on generative AI to make sense of data even faster. Now our customers in asset management, wealth management and insurance, and other data-intensive fields, can rapidly test hypotheses, spot trends, pinpoint risks, and capitalize on opportunities they may have otherwise missed.
With EMERGE, we did not go it alone. We have formed partnerships with AI pioneers like Anthropic and AWS, NVIDIA, OpenAI, and Microsoft, to remain at the frontier of what is possible for the enterprise. Together, we are committed to placing the power of data into the hands of every business so they can serve their customers at a higher level. The future of business will be data-driven and generative AI-powered. We are proud to be charting the course forward.
Hortz: Please help us understand some of the new terminology in this technology. I see the term “large language model” or LLM everywhere in reference to generative AI. Can you explain what exactly that refers to and what piece of the tech puzzle it represents?
Stevenson: A large language model is a very large AI model that is trained on data from the internet: things like Wikipedia, huge amounts of websites, but also huge arrays of journalism articles. Think of all the content articles that are written by banks, by researchers, by FinTechs, or by specialists in different domains. The LLM model is trained on all of this data. What it is able to do over time is to understand concepts from this data and piece them together.
When you send a prompt to chatGPT, it is going into this large language model. The LLM is interpreting your question or your prompt, and then it is predicting what the right answer should be based on what it has been trained on. And so, this is what could lead to “hallucinations” - where the LLM thinks it is doing a good job, but some of what it has put together might not actually be accurate or true. And so that is one of the problems.
There is a new way of training these large language models though which focuses on logic with steps to get to answers otherwise known as “planning”. In the AI world, planning is all about breaking down the question, having the AI train on and share the steps that it will take in order to get to the answer that you want, and if you agree with the logic, it will proceed to give you the answer based on that plan. We have seen industry luminaries such as Yann LeCunn, Godfather of AI advocate for planning as the next step in AI’s evolution in May 2023.
That is coming sometime next year in 2024. It will dramatically improve the quality of the answers and that will give us new capabilities in reasoning and interpretation of data and also the explanation of data. 2024 is going to be an extremely exciting year.
Hortz: Can you go more deeply into what exactly is the difference between AI and generative AI or an AI-driven platform versus a Generative AI-driven platform like your newly launched EMERGE platform?
Stevenson: That is a great question. AI really mimics the human senses. So, you can think of seeing, talking, translating, reading, those are really some of the core capabilities of AI and there are a number of different use cases in each of those categories. Now most of those are focused on identification and really what it typically involves is training, for example, on a series of images of dogs and cats. The system will eventually learn that certain representations are cats and others are dogs. The same goes for languages for audio. It is basically building up an understanding of the data.
What is different about generative AI is that, up until now, up until 2022 and 2023, most of the capabilities were focused on identification and then doing something with that information. Generative AI really removes a lot of those original barriers to entry. One, you do not need to train the data. A large language model is trained effectively on the whole internet and pouring in other data sets. Otherwise without LLM, it can cost millions of dollars, tens of millions of dollars actually to fully train this new model.
Think of it as pre-trained and in fact that is where the term GPT comes from - Generative Pre-trained Transformer. Transformer simply means that you can ask it any type of question and it will predict the next word. And in terms of predicting the next word, it is going to help to create an answer for you that is a good answer. So generative AI capabilities are all about creating something that is not there. In this case it could be, for example, creating an email or a video script or an image. And so, think of it as trained on a great deal of data already, and now, when you prompt it, it has the ability to be able to create something for you.
Hortz: How do you actually apply this generative AI technology to financial firms? As an example, what kind of advantages or unique capabilities have you developed specifically for financial services companies?
Stevenson: What we focused on through building the ForwardLane platform is creating insights. A large part of that is bringing together financial services companies with its data, such as data transactions, data packs, market data, CRM, perhaps even Morningstar ratings, and other information like that, as well as enterprise data which could be marketing campaigns, email opens, and other data that is being tracked. Our platform consolidates all that extensive data, links it together, and we have built tools to enable our clients to create insights. The insight really, or signal, is looking for something in that data, flagging it up, and then ranking, scoring, and prioritizing it.
We have spent a substantial amount of time with asset managers and wholesalers. We have done the same exercise with wealth managers and insurance brokers. What this has given us is a good idea of what kinds of information they would like and need when they are engaging with prospects and clients. We automated the analysis for them with these insights. We ran the data processing, scored, and prioritized the results, and delivered them into the workflow inside of Salesforce, for instance.
Now where our new Generative AI EMERGE platform comes in is we have taken all of our knowledge, all of the questions and queries and requests we have had from asset managers, wealth managers, and insurance firms and we ask, Can we build a system to answer 90 to 95% of these questions without us having to be there? And that is what emerges from our new platform. It is a very fast way to create insights out of your own data through a point and click interface. And then it has our Emerge GPT, which is the built-in Q & A capabilities to use the data that has come through the ForwardLane platform, in new and innovative ways. One example is asking questions of your book of business like, so where are my top opportunities?
Instead of having to dig through dashboards from the data science team, look through many different files to get to that, generative AI consolidates all of the data into one answer from your book of business with all the analytics and all the data that your organization already has run and generated. It is applying a level of understanding to it to help identify opportunities for you. One of the questions I like is, what have I missed? This taps into some of the generative capabilities, which can help with things like risk detection. I think of it as digging into hidden risks - blind spots - and insights which may not be obvious.
Hortz: How does the Generative AI platform also help create an action plan for the insights it discovers?
Stevenson: Once you have all those insights from your data, you can then interact with all those opportunities by asking the generative AI platform further questions and generate action steps: Why is this an opportunity? Can you give more information about it? What is the supporting evidence behind it? Then you can say, Okay, great to move forward with this client based on this information. Now, I would like to figure out how to engage with the client by asking, create an engagement plan for me, and then go ahead and write the first communication with the client in a personalized way which is related to what they are interested in.
We think that this is a fast and interesting way of communicating in a lightweight fashion that is more personalized with clients. In wealth and asset management, the use case is similar. We can empower you to review a very wide range of prospects in your territories and look at commonalities in the data and segment who you should be reaching out to, what to be discussing, and provide guidance as to what types of products to recommend. This ability to quickly process complex data from different sources in a simple ChatGPT style interface is a game changer.
Hortz: What other capabilities does generative AI bring to financial services?
Stevenson: Reporting is one area in which we recognized a pain point faced by managers and junior staff at all levels: the constant scramble to find data buried deep in spreadsheets, to answer urgent questions from leadership. It is a time sink that stifles productivity. But what if you could get detailed, accurate answers to any business question in seconds rather than hours?
That possibility inspired our latest innovation. We have integrated cutting-edge generative AI to enable real-time data queries and automated reporting. Now managers can simply ask questions in plain language to instantly generate insights from vast datasets. We are talking MBA-level answers in seconds. And the ability to push these capabilities out to senior executives so that leaders and their staff can self-serve the information they need to guide critical decisions.
The impact on efficiency and speed is revolutionary. We are seeing gains in productivity and responsiveness across the organizations we work with. And as these AI models grow even more accurate, the business potential is exponentially greater in terms of identifying risks, seizing opportunities, and connecting strategy to execution.
Of course, with the immense power of AI comes great responsibility. We have partnered with leaders in data security to ensure our solutions meet the strict requirements of the financial sector. With EMERGE, we are committed to developing generative AI that businesses can trust for ease-of-use, accuracy, and safety. We believe this technology marks the next chapter in data democratization and we are proud to be pioneers shaping its responsible and ethical integration.
Related: How Do You Read and Diagnose the Health of the Market?