Imagine an ecosystem in which one of the main leads that individuals need to make decisions is missing. Or, well, maybe not missing but scattered and distributed in so many pieces that putting it together takes an immense amount of time and effort, eroding the possibilities to actually act in favour of your interests. 

Más Voces developed elTOQUE Markets as an AI-powered solution to help stakeholders from within and outside Cuba understand the island's complex informal economy, particularly currency rates - a key challenge for readers of their main project, elTOQUE. This subscription-based data platform provides structured, real-time information on exchange rates, remittance costs, and food prices sourced from fragmented and unregulated channels, transforming what was initially identified as a content gap into a new potential revenue stream.

Operating in one of the most opaque information environments in the region, the project leverages artificial intelligence to address a fundamental lack of economic transparency while also testing a path to financial sustainability for a newsroom in exile.

The problem: Why elTOQUE Markets was born

In Cuba, the informal economy isn’t a peripheral phenomenon: it’s a foundational system that structures the interactions at every level. Due to the government's refusal to publish transparent economic data and the near-total absence of a functioning formal exchange market, economic actors depend on informal systems to survive and understand which is the most efficient current exchange rate. 

From currency exchange and remittances to food distribution, unregulated systems underpin daily life. But reliable, real-time information about these markets is scattered, unstructured, and hard to access, especially for entrepreneurs, diaspora investors, or anyone trying to make informed decisions.

Building the solution: From tracker to product

The roots of elTOQUE Markets lie in the outlet’s earlier project: a daily informal exchange rate tracker. Building on that success, the team designed a more comprehensive platform aimed at a wider range of economic indicators, with the potential to be monetised.

The project began with audience and market research to validate user needs, followed by the design of the platform’s information architecture. With this foundation, the team developed a MVP in the form of a paid newsletter, while also building the CRM and integrating payment systems to support its operations. These steps laid the groundwork for scaling toward the launch of the full web platform.

Tools and technologies

To collect and process the unstructured data behind Cuba’s informal economy, the team combined:

• Data capture bots deployed in WhatsApp, Telegram, and Messenger groups

• A NLTK-based model to extract and clean raw textual data on currency exchange offers

• LLMs like ChatGPT to interpret slang, typos, and evolving informal terms

• A CRM system to manage subscriptions and users

• MongoDB with separate collections to organise exchange data, product listings, and remittance info

• And a global vector database on top of MongoDB that is used for semantic searching and unify structured and unstructured data via enhanced microservice APIs

The team behind the platform

The team is spread across Cuba, Spain, Chile, Ecuador, and the USA, working under conditions shaped by exile and funding instability. Key roles include a product manager, a lead developer, AI engineers and data specialists, and CRM and frontend/backend developers.

Challenges faced

• Structuring the unstructured: Cuba’s informal markets exist primarily in private or semi-private digital spaces, like Telegram and Facebook groups. The language used in these spaces is informal, inconsistent, and constantly evolving - requiring ongoing fine-tuning of AI models.

  It also involves a great deal of normalisation efforts to uniform concepts: “There’s no universal way people spell the word ‘Zelle’ in these groups. It might be ‘cel’, ‘celle’, ‘celly’, or something else entirely,” explains José Jasán Nieves, executive director.

  The team also had to adapt to frequent changes in where and how economic information appears, as groups open, close, or shift platforms.

• Payment restrictions due to the embargo: USA sanctions block financial platforms like Stripe and PayPal from processing Cuba-related transactions. To bypass this, the team relied on QvaPay, a peer-to-peer cryptocurrency marketplace, which allows users in Cuba to purchase prepaid cards using local methods - a workaround that remains fragile.

• Remittance data still requires human input: Because many remittance providers operate informally - often just via a phone number or chat group - their data is inconsistent. The team still relies on some manual inputs for these sources, later structured by AI.

• Operating from exile: Developing this tool while in exile brings challenges around coordination, safety, and impact. Despite limited resources and the loss of half the team due to funding freezes, the project continued. Nieves assures: “This experiment has kept us going. It gave us hope when we had almost nothing left.”

The opportunities and learnings

The project offers a replicable model for data in opaque economies, showing how methods used in Cuba, such as collecting unstructured data from messaging apps and social networks and structuring it with AI, can be adapted to other countries with large informal markets, like Venezuela or Argentina, to generate actionable economic information. 

At the same time, it represents a new mindset for exiled media. For Más Voces, the initiative is not just a tool but a shift in sustainability strategy: “Most exile media rely on grants. But funding is drying up. If we can make this work, it could be our way to survive.”

Lessons for newsrooms

• AI can unlock public value in opaque environments: By structuring unregulated data, newsrooms can inform both local and international actors - from businesses to policymakers.

• Tech solutions must adapt to local languages and practices: Slang, typos, and informal channels challenge standard AI tools. Continuous training and human oversight are critical.

• Media innovation must include sustainability: elTOQUE Markets is as much a product experiment as a business one. It shows how media can build revenue by solving real user needs - even in restrictive contexts.