Last year, I made a set of predictions for data and analytics trends for 2025: “The Future of Data & Analytics: Trends to Look For.” Now that the year has wrapped, it’s time to put those forecasts to the test and grade how well they held up. 

Here’s my self-review, prediction by prediction.

Prediction: Semantic Layers Become an Essential Component for GenAI

Original Forecast:
Large Language Models (LLMs) are trained on a vast corpus of generally available data, making them useful for various tasks. However, LLMs lack insight into an enterprise’s proprietary business processes and terminology. For LLMs to empower GenAI for business, LLMs need business context.

What happened:
In 2025, semantic layers clearly emerged as a foundational requirement for GenAI in analytics, providing the trusted context, governance, and consistency needed to move AI from impressive demos to reliable, enterprise-grade decision making. 

Grade: A

I was right on with this prediction. Major platform vendors like Snowflake and Databricks introduced and expanded their own native semantic layer offerings, reinforcing the idea that semantics are now table stakes for GenAI-powered analytics.

Prediction: Data Democratization Empowers the Workforce Using GenAI

Original Forecast:
Self-service analytics platforms will empower employees at all levels to leverage data without requiring technical expertise. Tools with intuitive interfaces, augmented with AI and natural language processing, will allow users to ask questions and receive insights in plain language.

What happened:
While GenAI meaningfully expanded access to data through natural language interfaces in 2025, true data democratization fell short of expectations, as issues around trust, governance, semantic consistency, and organizational readiness continued to limit broad, self-service adoption beyond core data teams. 

Grade: C

I called this transformation too early. Enterprises need to deploy semantic layer platforms first to address trust, accuracy, and governance issues before leveraging GenAI for BI. For our customers who have already deployed the AtScale semantic layer, their data governance teams were comfortable using GenAI chatbots alongside their BI tools. However, for those who hadn’t yet implemented semantic models, their data governance teams blocked access to chatbots beyond simple proof-of-concept implementations.

Prediction: Business Intelligence (BI) Tools Get Smarter with GenAI

Original Forecast:
I predicted that 2025 would be the year we see true blending of GenAI and BI — not just productivity add-ons, but AI working behind the scenes to surface insights and anomalies that humans might miss.

What happened:
Generative AI has indeed become central to analytics tooling. Vendors like Snowflake, Databricks, and others are integrating AI agents and copilots into workflows. Gartner data suggests that AI agents will augment or automate a growing share of business decisions and that semantics (and strong underlying models) will be key to accuracy.

Grade: A-
The direction was correct. AI has become embedded in BI tools as vendor-specific copilots, but the breadth and depth of its impact on everyday decision-making are still evolving. For example, while Microsoft invested heavily in embedding Copilot across its ecosystem, 2025 revealed that Copilot adoption and business impact fell short of early expectations, prompting recalibrated targets and underscoring the challenge of turning AI innovation into widespread, reliable enterprise value. For our customers using Power BI on AtScale, we saw little adoption of Copilot beyond curiosity in the feature. There’s still more to be done here.

Final Grade: B+

Looking back on the year, my core predictions about AI’s integration, the importance of semantics, and disruption in analytics were directionally accurate and, in some cases, ahead of the curve. A few forecasts leaned optimistic on adoption velocity, but overall, the thesis held firm.

Key takeaway for 2026:
The future still belongs to analytics platforms that combine AI with governance, context, and performance and that help organizations do more than automate tasks: they enable better decisions. Check out my predictions for 2026.