Since our founding in 2016, we have been part of the unprecedented transformation in the world of bots and virtual assistants. We work to transform them into fundamental tools in the digital era, to improve the interaction between brands and their users, and we enhance their power as allies of organizations to reduce costs.
As part of our experience working with bots and AI engines, we chose to tell our story (and the capabilities of our products) from the concept of generations. We believe this idea reflects our evolution from simple rule-based bots to virtual assistants that with artificial intelligence (AI) are able to understand and simulate human behavior.
Below are our generations and what each one includes.
First generation: rule-based bots
These bots operate with a predefined set of rules (they do not learn or adapt). Their interfaces are simple and based on option menus. They integrate with company information systems. They receive images or voice messages and extract their content.
Some typical functionalities are:
- Customer service for frequently asked questions.
- Reservation of appointments for medical consultations.
- For utilities, notification of due dates, management of utility payments, schedules and availability.
- For online stores, sending abandoned cart messages, customer status change, sale completed and other events.
- Sending promotional messages and surveys via WhatsApp and other channels.
- Receive voice claims reports from insurance companies (e.g., a customer sends an audio report of what happened).
- Reading and processing of documents, such as medical prescriptions.
They are still in place and are sufficient for several organizations. These bots often function as the gateway to the world of virtual assistants, and companies tend to move from these to more sophisticated ones (next generations).
Second generation: hybrid (tree + basic AI)
Moving towards more fluid interaction, this generation combines basic rules with AI elements. Platforms such as Dialogflow and Watson enable these bots to handle more complex conversations and detect intent from keywords.
Second generation bots help to:
- Prequalify leads in sales.
- Replicate online stores in WhatsApp, with the possibility to display catalog and search for products.
- Recommend products or services based on preferences.
- Redirect the customer to a certain branch of inquiry, for example, to see options for a certain type of product or service.
Third generation: conversational (non-generative AI)
These bots are trained on documents (PDF, Excel, etc.) and websites. They look up the answer to the user's question or comment in the knowledge base and bring it back as they find it in that source. They do not use generative AI, so they always answer in the same way. They are ideal for when the company has few documents or sites.
Some of the new features of this generation are:
- Assistance in the incorporation and training of employees.
- For event centers, answering specific inquiries within a generality of items, for example, informing the client about the shows of the month and answering inquiries about a particular event, such as ticket prices and where to get them or admission requirements.
Fourth generation: generative virtual assistants
These bots are based on generative AI to perform complex tasks and learn in an advanced way. In addition, the personality with which the bot responds can be defined: from language and tone of voice to imitating a character. They are trained to work with and process large volumes of data.
We identified three types:
| Based on documents and websites | With system integrations | Mixed: documents and sites + integrations |
|---|---|---|
| Generative bot that is trained on documents and websites (such as the company's public website or an intranet). It searches the knowledge base for the question and generates unique responses by incorporating customer greetings and tone of voice, as well as taking into account the context of the conversation. | They integrate with customer systems (such as CRM, core banking, help desk, collection systems, etc.) to automate different use cases. For example: requesting a loan, scheduling an appointment, balance inquiry or account statement. | Documents and sites + integrations: combines types 1 and 2. It is our most powerful bot. |
Some of its most surprising uses are:
- Analysis of judicial dispositions and subsequent actions, such as freezing of bank accounts.
- Internal assistance to customer service agents, with integration to the company's systems.
- Travel reservation management that presents different options according to certain customer queries.
- Collection and delinquency management.
