Beyond buzzwords: types of AI in revenue tech

There are a ton of super exciting AI tools being built today for revenue teams - many of you probably work at companies that are building some of these! Yet, with all these tools in play, it’s hard to discern what’s actually new and what’s not. We always hear revenue professionals wondering what all these buzzwords actually mean - and most importantly, how can these tools help drive more revenue at my company?

In this article, we’ll be diving into the tech behind the buzz words, and discussing different types of artificial intelligence approaches that exist today, how new they really are, and how they can fit into your revenue stack to drive more revenue.

Machine Learning (ML)

At its core, Machine Learning is a method of data analysis. It involves the construction of algorithms that enable a system to learn from and make decisions based on data. This eliminates the need for explicit programming for each new situation, making ML a flexible and powerful tool for businesses.

ML has been around for a while and companies like Gong, Correlated, Salesforce, and other enablement tools have integrated it into their platforms to help teams make smarter decisions on large amounts of data.

While many of these companies are getting hot again since AI is “hot”, the underlying technology they’re building off of is rarely novel, but their integrations have been getting exponentially better driven by advances in the commercialization and ease-of-use of some of the underlying tech.

In the revenue stack, ML algorithms are commonly used in Customer Relationship Management (CRM) systems like Salesforce or Hubspot, where predictive analysis and customer segmentation can significantly enhance sales efficiency.

Supervised and Unsupervised Learning

Supervised and Unsupervised Learning are two subsets of Machine Learning that refer to the type of data used for learning.

In Supervised Learning, algorithms are trained on labeled data. This is useful if you already have a lot of historical data, like previously closed or churned opportunities, and want to analyze it. In the revenue stack, supervised learning can help in lead scoring, customer churn prediction, and more.

Unsupervised Learning, on the other hand, uses unlabeled data. The algorithm identifies patterns and relationships within the data on its own. This is helpful for discovering unknown patterns in data, such as new market segments or product expansion opportunities from customer calls.

Deep Learning

Deep Learning is a subfield of machine learning that uses neural networks with multiple layers - or 'depth' - to make sense of data. While deep learning algorithms are incredibly powerful, for revenue teams this technology often manifests itself in natural language processing (more on this in the next section!) or speech recognition.

In the revenue stack, deep learning is frequently used in customer service chatbots (like Intercom or Drift) or sales intelligence software (like Gong or Chorus), helping to automate and optimize customer interactions.

Natural Language Processing (NLP)

NLP is a branch of AI that deals with the interaction between computers and humans using natural language. The goal is to enable computers to understand, interpret, and generate human language in a valuable way.

In the context of the revenue stack, NLP can be found in sentiment analysis tools that gauge customer satisfaction based on their written or spoken words. This is particularly useful in social listening tools or customer support ticket analysis.

While not an incredibly new field, NLP algorithms have been a standard way to inject and analyze large volumes of text, but recently have fallen to the wayside compared to LLMs which often offer “smarter” more flexible ways to analyze text.

Large Language Models (LLMS)

Large Language Models are the hottest type of AI technology today. LLM models like OpenAI's GPT-3 and its successors are built on a deep learning architecture known as Transformer, these models have the ability to generate human-like text based on the input they receive.

Out of all the AI techniques discussed in this article, LLMs have had the most advancement over the last year, and as such a plethora of companies have sprung up to commercialize applications with techniques that are truly novel.

LLMs are trained on a vast corpus of text from the internet, which allows them to 'learn' patterns in human language and apply this knowledge in a variety of applications. This makes them great for things like drafting emails (Lavender, Apollo, Clay, and others are doing this), providing real-time support with better chatbots (i.e. Intercom), or analyzing feedback and other unstructured text.

Despite their impressively fluent output, their performance is highly variable and they do not 'understand' text in the same way humans do, and hence, their responses must be treated with an appropriate level of skepticism and oversight.

Conclusion

While LLMs are the hot new kid on the block of AI technology, it’s important to remember it’s less-flashy cousins like ML models have been proven and effective methods for helping revenue teams find insights and make predictions from large amounts of data. At the end of the day, the goal of adding AI to your revenue stack is to find ways that you can drive real revenue creation with best-in-class tooling.

Would love to hear what tools other folks are using — we’re always happy to chat AI tooling so definitely hit us up on Linkedin!