As the generative AI gold rush continues, investors need to be able to separate fact from fiction

Photo by Sanket Mishra on Unsplash

The generative AI ecosystem for enterprises is growing exponentially, with organisations like Salesforce launching their own AI-powered tools to rival recent big announcements from Microsoft, Meta, Google, and Baidu.

But the rush to roll out AI for fear of being left behind can come at a financial and reputational cost. Just ask Baidu, which saw its stock dive after a disappointing demo of its Ernie AI chatbot earlier this year. The underlying technology is revolutionary, but as with any innovation, the strategy for implementation is as important as the technology itself.

For investors, it can be hard to delineate between the genuine and the not-so-genuine regarding generative AI. Generative AI is a broad term for a technology with a wide array of applications, but what are those applications, and how can you tell who the real deal is?

Identifying use-cases

An excellent place to start is looking at where and how generative AI is being used.

The core technology of generative AI is Natural Language Processing (NLP). NLP understands user intent and extracts other important information from an input. More advanced generative AI technology leverages proprietary deep learning models (such as’s own Automatic Semantic Understanding) to improve human language understanding, thereby reducing the chance of misunderstanding user intent. Large Language Models (LLMs) meanwhile, are a type of machine learning model that undertake NLP tasks. Whilst NLP programmes typically focus on immediate context, LLMs pull their responses from vast swathes of data and formulate coherent and relevant responses for the user.

Different organisations apply this underlying technology in several different ways, which are summarised below:

Text/Voice requests

  • Summarising: Identifying key information through keywords and named entity recognition
  • Translating: Accurate language translation using sentiment analysis
  • Generating: Topical content and editing based on natural language descriptions


  • Generating: From natural language descriptions
  • Debugging: Scanning existing code for errors


  • Generating: From natural language descriptions


  • Internal: Navigating knowledge base from natural language inputs
  • External: Providing accurate information from natural language inputs

Virtual agents

  • Voice-based
  • Text-based

Virtual agents are the intersection of the understanding and generative capabilities of generative AI. For example, some of these virtual agents use Natural Language Understanding (NLU) to decipher user intent and respond with accurate information in a digestible and human format.

Moreover, other tools like conversational AI can harness the benefits of generative AI and make it less prone to ‘hallucinations’ and more usable for enterprises. When looking at the various generative AI offerings, investors must understand what each platform offers, and how it is using generative AI to benefit the enterprise as these are the initiatives that will be of most value to investors.

Navigating the generative AI landscape

A generative AI platform is only as good as the data it’s trained on. Open AI’s ChatGPT, for instance, is trained on a fixed data set, so whilst its responses may be more fluent, it sacrifices up-to-date information and accuracy. On the other hand, Google’s Bard continuously draws data from across the internet, thus prioritising relevancy but sacrificing accuracy. Seeing a theme?

Meanwhile, enterprise-focused AI firms are building composite systems – leveraging the best bits of Large Language Models and Natural Language Understanding as required. These hybrid systems give enterprises the ability to combine proprietary intent management, context handling and dialogue management solutions with the raw processing power of LLM-enhanced tools – without the limits on LLM’s ability to offer consistently predictable outcomes.

Henry Vaage Iversen

Microsoft has recently announced Chat GPT-4 will be integrated into its flagship search engine Bing and the Microsoft Office 365 platform. However, for organisations keen to bring generative AI into their business processes, the problem of vendor lock-in is hard to ignore.

The announcements from Salesforce, Baidu and Snapchat are no doubt exciting for bringing the power of generative AI to a wider audience. Still, these are internal platforms built by these organisations to serve their own customer base. These announcements mean very little for businesses with the appetite to improve customer service but are lacking the AI knowledge base to create a bespoke solution to meet their needs. An open and competitive ecosystem, where AI offerings can be implemented with relatively minimal lead time, is critical to bringing the transformative power of this technology to all enterprises.

We need to see more players in the market that can support organisations with AI-powered virtual agents across their whole business rather than just across a single software platform or suite of applications before generative AI becomes commonplace in the enterprise.

Get your priorities in order

During this period of generative AI proliferation, the key challenge for the enterprise is identifying what generative AI applications can unlock the most value for their business and improve the customer experience. Often, this means prioritising accuracy and reliability above all else. Investors need to look at how AI firms are leveraging generative AI to build the most accurate and reliable systems and not be tempted in by unrealistic promises.

Building a generative AI strategy requires a clear roadmap for which solution would be the most effective for achieving a business’s specific goals. For investors, this means identifying the applications with the broadest appeal, and currently, that lies with virtual agents. Virtual agents work for any company where search, code or text summaries are often hyper-focused on specific industries. Moreover, virtual agents can work both externally for customer-facing businesses and internally to improve employee experiences.

What is clear is that not having a plan to implement generative AI is akin to rolling over and giving the competition a huge head start. We are at the start of a new era for business operations, and it is up to stakeholders to ensure they make the right choice. It’s clear that they need to be looking at generative AI and how it might assist their business, or else miss out on the benefits of this transformative technology.

Henry Vaage Iversen is CCO and Co-Founder of