How a team of researchers from MBZUAI is using AI to empower businesses with instant data analytics

A team of researchers at Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) has developed a new AI system that reimagines the role of the data scientist — making advanced analytics accessible, explainable, and lightning-fast for business leaders. 

Launched by Farkhad Akimov, master’s graduate in Machine Learning; Zangir Iklassov, Assistant Teaching Professor in Machine Learning;  and Munachiso Nwadike, Research Engineer, Data Wise is an autonomous AI platform that turns raw customer data into actionable and justifiable recommendations through AI agents that are specially built for reasoning. 

The platform is designed to be adaptable across industries and functions: “It is fundamentally intended to replicate how a data scientist operates in real life,” explains Nwadike. “Data scientists are often handed a pile of numbers, text, or images and asked to figure out what is going on and what to do next. Our platform mirrors that process, building predictive models that help businesses understand what will happen in the future.”

This makes the tool just as relevant for small start-ups without data teams, as for large enterprises that need rapid responses at scale. 

“With organizations today drowning in data, we wanted to build a system that empowers leaders to act without waiting on technical teams,” he adds. “Data Wise helps anyone – not just data scientists – test hypotheses, discover patterns, and make decisions, all in the same conversation.”

Farkhad Akimov

Meeting a regional need 

The timing is critical. With the volume of data exploding, access to skilled analysts remains a bottleneck. The U.S. faces a shortfall of more than 250,000 data scientist, with McKinsey projecting that by 2026, demand will exceed supply by roughly 50%. And in the UAE and wider Gulf, the challenge is even more acute. 

Data science ranks among the UAE’s top ten skill shortages, according to a report by Cooper Fitch. A joint study by Strategy& and LinkedIn the same year found that while GCC digital professionals often excel in soft and managerial skills, they lag in advanced technical expertise such as data science. In Dubai, the Digital Skills Study 2022-2023 reinforced this trend, identifying data and AI as two of the most significant workforce gaps. 

With the augmented analytics market projected to grow from $ 29.81 billion in 2025 to $102.78 billion by 2030, the demand for scalable intelligence has never been higher. 

Zangir Iklassov

How Data Wise works: Leveraging agents for end-to-end analysis 

Data Wise’s analysis is done in three main steps: 

• Users upload data such as purchase history, surveys, or support logs. 
• They then specify business goals, and the system generates insights. 
• Finally, users chat with Data Wise to refine and receive plain answers backed by model reasoning.

The platform generates its ideas, or “hypotheses,” to explain patterns in business data. It tests those hypotheses to ensure that only insights with the highest confidence are surfaced. To do this, the platform relies on a group of large language model (LLM) agents that work in parallel. 

“Instead of merely handing you a dashboard and expecting you to decode it yourself, Data Wise’s agents manage the full analytical process end-to-end,” says Akimov, whose thesis – developed under the supervision of Martin Takáč, Deputy Department Chair and Associate Professor of Machine Learning – was the basis for the platform’s development. 

Users can pose custom business questions and receive model-backed, easy-to-understand responses. Unlike similar platforms, Data Wise’s agents autonomously write their own code, run diagnostics, and request additional data when the inputs are too weak to support a solid conclusion. 

“The key limitation of most platforms – even great ones like Tableau or Vertex AI – is that they automate workflows for data scientists, not decision-makers,” explains Iklassov. “We built Data Wise so a business owner, not just a quant, could get to the ‘why’ behind the data.” 

“The core innovation lies in how the platform operationalizes data-driven reasoning,” adds Akimov. “Using LLM agents capable of executing Python and R code, the system can surface latent patterns, assess correlation versus causation, and even detect where additional data is needed. It delivers analysis at conversation speed, rather than after several days of iteration by a skilled team.”

Munachiso Nwadike

Putting the platform to the test 

To validate its performance, the team tested it against top-ranked specialists on Kaggle – the world’s leading data science competition platform – using datasets covering customer churn, health risks, diamond pricing, and used-car values. Across the board, the system outperformed expert-built solutions. 

The platform is now live as a working demo on AWS, showcasing its core functionality for testing and demonstrations. Recent updates include interactive dashboards and the ability to generate and download reports, making insights easier to explore and share. 

“We envision Data Wise unlocking value for decision-makers in every vertical – whether it is predicting patient readmission in hospitals, optimizing marketing budgets, or reducing inventory costs in manufacturing,” says Iklassov. “The goal is to make proactive strategy a default, not a luxury.” 

Support from IEC 

Data Wise is being incubated through MBZUAI’s Innovation and Entrepreneurship Center (IEC) – the University’s AI-native incubator that blends the technical substance of AI with business insights and an intimate understanding of the local market.  

The team has been awarded a grant and received mentorship, business support, and guidance on filing a provisional patent – all of which has been instrumental in shaping both the product and the startup’s direction. 

“We are continually looking to expand and improve our product,” explains Akimov. “We are building on our demo to create the next version, which not only understands enterprise data but adapts to each organization’s specific needs, securely, transparently, and with the ability to scale.” 

This vision aligns with the global trend toward explainable AI and autonomous decision support. Allied Market Research valued the explainable AI market at $6.2 billion in 2023, with projections to reach $39.6 billion by 2033. 

“We believe data science is moving from the world of static dashboards to dynamic conversations,” says Nwadike. “With generative agents that understand both code and real-world context, we are entering a new era where business leaders can interact directly with their data and trust the answers they receive.”