WhatsApp voice messages are the most convenient and the most hateful way to communicate when using this application. There is no middle ground. As senders we love them, as receivers we hate them.

LuzIA, the personal assistant integrated into WhatsApp, helps us translate the voice messages we receive and supports us in other tasks we can "play around" with. LuzIA, one of thousands of Generative AI solutions appearing on the market, has reached a record figure of 1 million users in less than 2 months. It is the fastest-growing Spanish application in our market. Instagram needed practically twice as long and Spotify slightly more, 5 months.

The most common comment among those of us who are following (or rather, chasing) the Generative AI phenomenon is that we don't have enough time in our lives to identify and test the huge amount of solutions appearing week after week. Everything indicates that we are facing a disruption of similar and even superior characteristics to what the emergence of the public internet or smartphones meant.

Welcome to the era of Generative AI, a new disruptive cycle has begun.

Microsoft and Google, in that order, are leading the advances among the incumbents, competing fiercely to incorporate new capabilities into their flagship products. Their communication campaigns are especially aggressive as, for the moment, these AI superpowers applied to their most emblematic products (MS 365, MS Dynamics, Google Search or G Suite) have not been launched on a massive scale and we will possibly have to wait until the end of the year. It is clear, however, that these new models are going to revolutionise two aspects that have been key until now in any business strategy:

• Consumers are going to stop browsing and will rely more and more on their AI assistants to get informed and make decisions, transforming current browsing and "conversation" on social networks into models of personal relationships with brands and their products. We are entering a new era of hyper-personalisation.

  
  

• Employees are going to have access to new tools with the capacity to generate huge increases in productivity, and their use will require the development of new skills and competences to respond faster and better to the new competitive variables of the environment. We are facing a new era of hyper-speed.

OpenAI, Microsoft's ally since 2019, continues its commitment to generating new models (GPT4 on the market and rumours of GPT5 by the end of the year) of high capabilities and high development and training costs. Its paid API strategy, which is basic to its monetisation and profitability model, is facilitating the emergence of new solutions (LuzIA uses OpenAI's APIs) that are being adopted by users at a breakneck pace, generating exponential growth dynamics that feed the ambition of investors, business owners and entrepreneurs to capture a slice of the pie.

Based on OpenAI's paid APIs, large conventional companies are investing time and resources to develop their own Generative AI solutions, although doubts about the confidentiality and privacy with which GPT treats these companies' sensitive information and the cost associated with using these APIs is proving to be a drag on these developments.

Microsoft and Google, but also META, Nvidia and other big tech companies are developing large models (Large Language Model, LLM) with trillions of parameters and high development and training costs to be the dominant ecosystems of Generative AI, in the same way that today iOS or Android are the dominant ecosystems for application development.

High-capacity LLMs are not the only path for the development of Generative AI; open-source models are proving to be faster, more customisable, more private and more capable in economic terms. These "Smart LLMs" only require a few billion parameters and a few weeks of training to become ecosystems on which to build solutions adapted to the specific needs of companies with greater guarantees of privacy and confidentiality. With a $100 investment in a few weeks of training and 13B parameters, "Smart LLMs" compete head-to-head with LLMs (GPT4, Google's PaLM, ...) that cost $10 million in investment, months of training and more than 500B parameters.

By the way, many of these "Smart LLMs" are developed on the open-source basis of LLaMA 13B, the basic model of META's LLM, which has oriented its strategy towards the massive community of open-source developers to win the battle of the dominant ecosystems.

A new era of productivity is upon us and our recommendation for businesses and their leaders is "KEEP CALM & EXPLORE".