Gartner has announced the main data and analytical trends that should shape the industry by 2025, highlighting various challenges that organizations may be confronted, including organizational and human problems.
“D&A goes from the field of some to ubiquity,” said Gareth Herschel, vice-president analyst at Gartner. “At the same time, D&A chiefs are under pressure so as not to do more with less, but to do much more with much more, and which can be even more difficult because the issues are in progress.
AI agents are in good place in these trends, offering solutions for ad hoc adaptive needs, flexible or complex. Gartner advises that beyond relying on large-language models, the D&A leaders should ensure that AI agents can transparently access and share data between applications.
The concept of agent analysis implies the automation of closed loop trade results using AI agents for data analysis. Gartner suggests piloting use cases that incorporate information into natural language interfaces and supplier roadmap assessment to integrate digital work applications. Establishing robust governance to minimize errors and assess data preparation through specific principles for AI is also recommended.
The models of small languages are highlighted as an alternative to large languages models to obtain more precise and contextually appropriate AI outputs in specific fields. Gartner recommends providing data for recovery, increased generation or refined personalized models, in particular for on -site parameters to manage sensitive data and reduce calculation costs.
Gartner advises an exploitation of composite AI, which combines several AI techniques to improve impact and reliability. D&A teams should go beyond the generative models of AI or large languages, using data science, automatic learning, knowledge graphics and optimization to create full AI solutions.
The emphasis on highly consumable data products highlights the importance of treating critical use cases and settling these products to mitigate the challenges in data delivery. Gartner highlights the importance of creating reusable minimum data products and viable compositions that can be improved in an iterative manner. He also underlines the need for consensus on performance indicators between production and consumption of teams to precisely measure the success of data products.
Another key trend of metadata is a key trend. Gartner suggests starting with technical metadata and extending to commercial metadata to provide an improved context. This approach facilitates the creation of data catalogs, data line and use cases led by AI by automated discovery and analysis.
The deployment of a multimodal data fabric is invited, capturing and analyzing metadata through the data pipeline to improve orchestration requests and improve operational excellence thanks to techniques such as dataops, while allowing the development of data products.
In the field of synthetic data, Gartner highlights the need to identify where data is missing or costly to obtain, because synthetic data can offer confidentiality alternatives to advance AI initiatives.
Decision intelligence platforms are encouraged as part of the transition from a vision focused on a vision centered on the decision. Priorifying urgent commercial decisions for modeling, the alignment establishment on DI practices and the evaluation of DI platforms is advised. The fight against ethics, legal and compliance in the automation of decisions is considered essential.