Top Three 2026 Tech Predictions for Data and Analytics

2025 Background

The rush to adopt AI crashed against poor data quality (DQ). As per Qlik: “77% of companies with $5B+ in revenue expect poor AI data quality to cause a major crisis” (“Data Quality Is Not Being Prioritized on AI Projects,” Qlik, 2025).

Very often, DQ problems were fixed by business intelligence (BI) teams while preparing data for dashboards/reports. Once BI teams were removed from the pipeline, DQ problems became accentuated by AI that changed the proverbial “garbage in, garbage out” into “garbage in, disaster out.” Poor data quality results not only in incorrect BI report numbers, but in incorrect strategic analysis and advice performed by AI.

AI-powered conversational interface has shown the drastic difference between getting answers instantaneously via natural language vs. via dashboards that take several weeks or even months to build and require training to use. BI dashboards are falling out of favor: “72% of business leaders say they aren’t satisfied with how long it takes to get answers from their analytics teams. By the time a new dashboard or report is built, the insight may come too late – the market has shifted or the opportunity has passed” (“How AI Killed The Dashboard,” Zive, 2025).

Microsoft introduced an ontology-based IQ layer (Foundry IQ, Fabric IQ, and Work IQ) that reintroduced “ontology” to the forefront of Agentic AI discussions. Ontology is a concept that used to be popular in early 2010s during active deliberations about using Semantic Web Standards for data harmonization and advanced analytics.

Predictions for 2026

1) AI will ensure data quality – much needed for AI

Every analytics platform will have data quality management (DQM) capabilities – most likely powered by a specially trained AI agent. Recent examples of vendors having added DQM include Dataiku, Alation, and Atlan. CluedIn, Reltio, Informatica, and Ataccama have had DQM as part of their core offerings for a while but recently added AI agents to run the capability. This trend will continue since more structured and unstructured content will be both generated and consumed by AI – then, AI itself (well, DQM-focused agents, to be exact) will take care of the data quality from record creation (unless inhibited by legacy technology) all the way to the information consumption (including the data augmentation processes). This does not mean that we can consider the DQM problem fixed forever – data quality is in the eyes of the beholder, and DQ requirements must still be established or verified/confirmed by human beings. Don’t rush to relieve data stewards from their DQ duties (see our blueprints on data governance and data quality management). Since data definitions are typically part of DQM, the latter will extend into the enterprise semantic integration (ESI) layer and become part thereof (see the “Ontology” section below).

2) Traditional business intelligence tools will start fading away

As predicted back in 2021, traditional static visualizations (reports, dashboards, slide-decks) are being replaced by AI-generated stories, voice-driven interactive information research, and animated visualizations that can change as the dialogue between the user and AI assistant progresses. Graphs and charts would still be required in certain visualization cases, but BI as a functional capability will change; i.e. data quality and data preparation will be mostly done by AI agents. Data analytics run by AI would not need complex SQL structures – data definitions and semantic integration would be all that AI needs to analyze the data. Visualizations – both still and animated – will be mostly generated by AI, which is rapidly maturing in this domain. Standalone BI tools, therefore, will become less attractive, while the ones that are tightly integrated with a large data management and analytics stack – serving as visualization tools for the stack components – will remain relevant. If you need to renovate your data & analytics platform, use Info-Tech’s blueprint Select Your Data Platform.

3) “Ontology” will become the buzzword of the year

Microsoft brought ontology into the spotlight by placing it at the core of its IQ layer. This means that we have conquered the “Data” and “Information” layers of DIKW pyramid and are moving Analytics to the level of Knowledge. Now more people will see the difference between property graphs and semantic (ontology-based) graphs – up until now, both standards have been used for knowledge graphs with the property graph dominating as it is much simpler to create and manage compared to ontology. The new wave of AI Agents requires not just semantic associations (something/someone is relevant to something/someone), but a more complex knowledge representation – including definition of conceptual classes (e.g. Program, Project) and their hierarchies (Program > Project), reification of properties and relationships (e.g. employment was valid from [date] to [date]), establishing constraints (e.g. fish can be edible only fresh or after being kept deeply frozen) and disjoint sets (e.g. voters for one party and voters for another/other parties). This is what ontology enables and property graphs struggle to represent. Precision and predictability of analytics will depend on the quality of the governing ontology. Organizations should expect new AI agents built and trained for creating and managing ontologies, but those may take some time to mature. Meanwhile, use data cataloging and master data management tools that can serialize their data/metadata as RDF/OWL (Semantic Web Standards). Alternatively, hire a professional ontologist – they are a very rare breed, however.

Thus, we should expect that 2026 will:

• Make the DQ problem less acute – but we will still need data stewards and SMEs.
• Change BI from the standalone data augmentation and information consumption technology to an integral component of a larger technology stack.
• Move the differentiating focus of analytics from the Data/Information layers into the Knowledge layer – i.e. they who have the best ontology can have the best analytics (all other technology factors being equal).

Want to Know More?

• Microsoft Adopts Ontology-Based IQ Layer for Agentic AI
• Advancing Analytics to the Level of Knowledge
• Solidatus Expands Metadata Governance With Agentic AI and Bi-Temporal
• CluedIn: Graph-Based MDM With Agentic Data Management