Chatbots are software that simulate human conversation. Programmed to receive and respond to messages, whether text or voice, automatically and instantly, this tool has been endowed with great sophistication with the contribution of Artificial Intelligence (AI), so much so that a specific term was coined to designate this technology: conversational intelligence, an asset for many companies wanting to improve the experience with consumers; and at the same time optimize processes, generating savings.
“We rarely see a technology that is a win-win for both the company and the end customer. For example, if it is possible to avoid making a call to make an appointment, make a complaint, or make a purchase; you avoid it, because you are committing 100% of your time to this service. The same thing happens for the company, the attendant needs to be 100% dedicated to you, they cannot provide more than one voice service at the same time. So, it is a cost reduction for the company and a much greater convenience for those being served”, assesses José Caodaglio, CEO of Colmeia – Meta’s official representative in Brazil to offer personalized communication solutions via digital channels between companies and consumers.
A reference for B2B companies in this field, the brand that was born in 2019 as a startup, today already has more than 100 customers and an investment of R$25 million by Crescera Capital to triple its size, showing the heated scenario of bots in the age of AI . According to the 2024 “State of AI Application Strategy Report” by F5 (global security and Multicloud applications company), 75% of organizations in the world are investing in incorporating AI into their processes, one of the most mentioned applications (36%) is in chatbots for customer support. Exponential growth is expected for this digital service market in the coming years.
Bots: from Turing to conversational intelligence
It was 1950 when mathematician and computer scientist Alan Turing decided to test the ability of a machine to simulate human thinking, and thus interact with people, without them discovering that it was a robot. If the simulation lasted for at least five minutes, the computer would pass what became known as the Turing Test, even if after this time, the discovery occurred. Turing’s “joke” paved the way in the 1960s for the first Natural Language Processing (NLP) software, ELIZA. The creation of MIT researcher Joseph Weizenbaum was basically to make computers understand, interpret and interact using human language.
In the 1970s and 1980s ELIZA was refined into two more advanced tools: PARRY and Jabberwacky, by scientists Kenneth Colby and Rollo Carpenter, respectively. However, it was only in 1995 with the launch of Richard Wallace’s Artificial Linguistic Internet Computer Entity (ALICE), that historically it was considered that there was a migration to an AI language, as the software, the first to be run on a computer , worked with AIML – Artificial Intelligence markup language.
For the millennial generation, it’s easy to recall what came later: SmarterChild, available on AOL IM and MSN Messenger, became extremely popular in the early 2000s, when users were able to get answers to questions such as game results, weather forecasts, weather, cinema programming… And from then on the launches never stopped: Siri (Apple), Google Now, Alexa (Amazon).
Today, different programs exhibit rapid interactivity, are able to provide information in real time, in addition to performing very specific functions. A bot can, for example, automate repetitive tasks such as collecting feedback on medical appointments, or registering participants for an event. Everything will depend on learning and the technology embedded in each solution developed for different companies and market niches.
Reshaping business
And now, in times of LLMs where the Turing Test has to be improved to detect increasingly subtle nuances?
The field is fertile: Brazil has more than two digital devices per inhabitant, according to Fundação Getúlio Vargas (FGV); it is the second country in the world with the highest average time online, more than nine hours of connection, according to research by We Are Social (communications company) and Meltwater (online media monitoring company); and more than 90% of the population says they have already used some type of virtual assistant, according to a study by Ilumeo Data Science Company.
It is precisely looking at this panorama that many companies are seeking to invest in conversational technology with clear strategies: improving the customer experience, expanding the number of consumers and optimizing team work.
“When we talk about conversational intelligence, the first use case we think of is customer service, that is, the customer contacts the company with the purpose of being served. But it’s much more than that. Conversational intelligence has to be active, part of companies’ primary efforts, which logically is to increase their sales. In this sense, the cycle of investing in Performance Marketing (or making active calls), directing the audience to an intelligent Bot, measuring performance in real time and feeding back to algorithms (Meta or Google) is vital for companies that aim for maximum profitability. and efficiency. What about when your company is engaging 50,000, 100,000 people in a campaign? Controlling engagement, recency, steps of the communication rule are no longer optional items. Some say that working on this scale is expensive, but without a doubt, not reaching and providing the correct experience for your customer is expensive”, explains José Caodaglio.
According to the executive, the advent of ChatGPT naturally led to a rush by virtually all companies to use the technology in their processes, whatever they were. However, the executive notes that LLMs, powerful as they are, will end up depending on the same maturity that many companies are still struggling to achieve.
Technology companies like Colmeia that develop solutions that allow companies to abstract the high complexities of using various AI strategies, communication channel providers, telecommunications operators, ad providers, rules and regulations such as LGPD, Analytics, CRMs among others, provide a great shortcut for implementing conversational technologies:
“There are a lot of moving parts. We saw that companies did not find it easy to prospect information in-house, they thought that AI would be a shortcut, but they realized that they needed to do some homework to deliver this information in a relevant and personalized way. Because this information comes from the company’s own internal, classified and sensitive data, if the company cannot deliver this to Artificial Intelligence, no matter what intelligence you have, it is not possible to work with it. There is still a barrier for companies to be able to organize their own data and we are able to offer more optimized ways to obtain quick results without having to reevaluate the company’s entire footprint”, says the CEO.
Another relevant factor is delivery time. “The nature of digital operations is agility. The ability to test hypotheses quickly, make mistakes and correct journeys in minutes with autonomy and without entering into complex flows like those of more traditional technologies is imperative.” The average implementation time for sophisticated journeys, says Caodaglio, is three days, even if they are extremely personalized. With a portfolio of clients that ranges from the services sector to the demanding financial market, and delivering solutions that range from registrations to journeys with biometrics, the executive believes that conversational solutions will make up a key portion of the segment that was only qualified as e-commerce. “The results are quickly noticeable, Brazilians really identify with conversational channels”
What’s to come
If at a certain point in the evolution of chatbots, it seemed impossible for software to be able to overcome the five minutes of the Turing Test, with the evolution of Artificial Intelligence and the sophistication of programs, the expectation is that more and more interactions will become very similar to human interactions. . However, the doubt of whether they are being served by the machine or by the human is already an issue that has been overcome, since the companies’ focus is on the value generated in this interaction.
Still, the bet is that these tools will expand humanization and personalization in the coming years. These channels must be increasingly sought after to replace traditional communication and the greater use is directly proportional to improvements in speed, analytical capacity and accuracy of automatic responses. At the other end, this technological migration becomes mandatory for companies, as a condition for remaining competitive.
José Caodaglio analyzes that the future of chatbots is related to investment, research and development of new functionalities, so for companies that offer these solutions it is essential to always be one step ahead of market needs. “We are entering several new markets, in marketing performance, we operate digital marketing, digital campaigns very well and we will continue to make innovations in Artificial Intelligence. Our AI is multi-strategy, which means that we can put Google Gemini working together with ChatGPT and Meta LLama, along with more traditional technologies like machine learning, all combined to interpret our customers’ inputs. These intelligences exchange information with each other, and with this we are able to interpret people’s needs much more effectively. We observe that there is a lot of room for investment in technology, there is a huge gap between the technology that large providers offer and the technology that companies need to use. There is a lot of room for innovation and research to reduce costs and increase efficiency for companies. We have applied for a patent in Artificial Intelligence, precisely to solve specific and practical problems that companies are often not even looking at yet. So, there is a lot of room for evolution.”
(fonte: MIT Technology Review)
