Generative AI “is poised to deliver the most significant economic boost and job change since agricultural and industrial reforms,” says Alberto García Arrieta, managing director and Head of Data & AI at Accenture for Spain and Portugal. “Most organizations see it as a path to greater innovation and revenue growth, and not so much to cost reduction.” The problem? Its high initial outlay can open a gap between large corporations, which can afford it, and SMEs that cannot. At this point, open source models (programming instructions or source code in the public domain for use, distribution or modification) can open the door, or at least open a window, to the triple opportunity that Accenture predicts: acceleration of economic value, boosting business growth and more creative and meaningful work for people.
“Can’t a small company build business with the big proprietary models like GPT? Of course it can. But the open source “It allows a series of possibilities that make it easier to develop service through it,” says Miguel Lucas, Global Director of Innovation at LLYC. The entrepreneur downloads it, integrates it into his infrastructure and can adapt it. In addition, the solutions no code (without code) and low code (customizable and with minimal manual coding) make it easy, without programming knowledge, to develop simple applications by overcoming the barrier of technical knowledge. “They make the implementation of AI more accessible, in terms of cost and time. Its cost is amortized in one year,” explains Íñigo Zúñiga, founder of the agency marketing digital Media Code.
Whether it is universal also depends on the small ones
Global adoption of AI could unlock up to $10.3 trillion in additional economic value by 2038, Accenture estimates. But first it should be global: organizations of any size should use it on a large scale across all industries, sectors and value chains. It should be remembered here that SMEs generate just over 62% of both Gross Added Value (GVA) and employment in Spain.
When asking artist, professor and director of Digital Transformation at VML/The Cocktail Alberto Barreiro about the impact of generative AI on small businesses, he referred the question to Joy, an AI with whom he talks about current issues. She concluded that its impact focuses on automation and efficiency, creativity, innovation, personalization and customer experience. For Barreiro, it can already close the budget gap that separates SMEs and corporations on issues such as “marketing digital and content creation, customer service or data analysis.”
Generative AI is set to deliver the most significant economic boost and job change since agricultural and industrial reforms
Alberto García Arrieta, Managing Director and Head of Data & AI at Accenture in Spain and Portugal
Together with its technological partner Danngos, Código Media is already tackling AI projects for environments as diverse as accounting or agri-food, for example for the Aragonese company Exafan (around 120 employees), a manufacturer and exporter of agricultural equipment. “We implement solutions that automate inventory management, invoice management, after-sales services with chatbots and automated generation of budgets… In this way we optimize work time and reduce delivery time to the client,” explains Jorge Dannenberg, CEO of Danngos and Technology Director at Código Media.
Creativity and empathy ‘made in SMEs’
“You don’t have to throw away what you have,” explains Zúñiga, because the new tools are integrated with existing ones in companies that cannot, or do not want to, afford to completely renew their technological wardrobe. The most interesting thing about this model? “It takes a look at everything that has happened and generates proposals for improvement that, obviously, must be validated by a person.” That is why Dannenberg insists that they do not come to replace but to add, although they require organizational changes and the retraining of the staff. For Zúñiga, creativity and empathy in SMEs can be the differentiating element. “AI is very similar to itself; it tends to always do the same thing if it does not receive the right instructions and data.” “It is like a gifted child: it is very intelligent, but it needs a guide to know which way to go.”
Barreiro agrees on the disruptive potential of accessible AI as an amplifier of human capabilities and a co-pilot for companies that were previously unaware of deep technological transformations. He argues that it will not only help them do the same thing in a more optimal way, but also transform the business model, innovate processes and inspire ideas. For example, it will allow a pharmacy to notify a user, via mobile phone, that she needs to take an osteoporosis pill on an empty stomach. Or restaurants will offer an uninterrupted reservation service in any language.
In the future, any local business will be able to access databases with types of users and customers.
Alberto Barreiro, artist, professor and director of Transformation at VML/The Cocktail
In this human impact on AI, and vice versa, Barreiro is fascinated by synthetic archetypes. They are generalisations of profiles used to explore concerns, needs or new demands of a certain segment of the population. Take for example Gabriela, 32 years old, a sustainable fashion entrepreneur from Brazil, who can be consulted about a banking product before its launch, for example. Synthetic archetypes, like other recent innovations, are currently only available to large companies, but will gradually spread to SMEs: “In the future, any neighbourhood business will be able to access databases with user and customer typologies,” Barreiro predicts.
It is better to train with real projects and problems
Even if the costs of generative AI fall, they remain a barrier to entry for SMEs. That is why it is key to raise awareness and train the team in open source and self-programming skills, in “how to interact with the main platforms or analyze and identify the problem and the formulation of questions well through data engineering.” prompts”, adds Carla Noguera, Account Director at Código Media and responsible for the training process. Training in tools no code and open source (for example automation platforms and applications like Zapier, UiPath or Dataiku, or software Open source code (such as PyTorch or TensorFlow) allows for the creation of specific solutions for SMEs without advanced programming and are accessible through workshops and online courses. The accessible philosophy of open source does not only concern the design of a softwarebut rather training to develop it. Noguera recommends practical learning through pilot projects so that employees can experience solving real problems.
We are therefore talking about “a great paradigm shift”, according to Barreiro, and therefore a deep, prior and inevitable cultural change, with doomsayers and optimists integrated. “17% of the managers we have interviewed anticipate an increase in their market share of 10% or more thanks to this technology”, says García Arrieta. Although it is not being implemented in an orderly manner, says Lucas, something that “would guarantee results by mitigating the risks”. “It is more difficult to change the mentality of the human teams at the head of companies than the technological challenge itself”, says Zúñiga. “I think the key is the CEO; if he and the management team are not clear about it, forget it”.
Cost/utility ratio
39% of Spanish SMEs use AI “from time to time” or “frequently”, and 62% believe they have a good or very good level of knowledge in this regard, according to a study by hosting provider Ionos. However, only 21% implement it, and almost one in four would do so if the costs were less than 1% of turnover. At the moment, 27% refuse to use it.
Most Spanish SMEs believe that AI streamlines and simplifies processes, but also that its impact is difficult to control and poses a risk of misinformation or data security. 71% advocate that its use be regulated by governments.
According to the same report, 74% of these companies consider it quite or very positive for customer service, image creation and word processing, compared to 63% of their counterparts in the US, 60% in France, 55% in Germany and 48% in the UK.
The discreet charm of small models
It should be noted that SMEs require AI applications to remain competitive against the size and budget of corporations. “But they cannot integrate them directly,” says Esteve Almirall, professor at Esade’s Department of Operations, Innovation and Data Sciences. In general, they tend to lack the capacity, experience and a workforce with the necessary training, he argues. But they will have, or should have, a whole ecosystem of startups and companies that can help them to achieve this. software that will customize open source AI models to adapt them to each specific need, be it a customer service, a simultaneous translation system or an invoicing assistant, to give just three examples. It is there, in these vertical blocks of specialist AI (niche market, a kind of size S with no possibility of scaling), where the expert sees the gap compared to the large proprietary models, where small companies can sneak in to provide themselves with tools that would otherwise not be affordable.
“Small open source models don’t compete with the big ones, they don’t need huge processors, and they work relatively well,” Almirall explains. It’s true that they are not as powerful as proprietary models (a GPT4 or a Gemini), but they can perfectly cover the requirements of an SME. “We don’t need a GPT for everything,” he stresses. It would be enough to customize the code of Llama de Meta, the solutions of the French startup Mistral AI or Claude de Anthropic. “Another advantage of open source is that it lowers prices” because it increases supply, stimulates competition and pushes down the rest of the market. But the Esade professor calls for the urgent creation of the aforementioned ecosystem of suppliers: “If we don’t help these companies to emerge now, we won’t do anything.”
