Business intelligence is going through a period of significant change. Advances in AI, growing demand for better self-serve analytics, and shifting expectations about how data fits into everyday work are reshaping how companies think about and use data.
In this post we outline key trends shaping BI and analytics in 2026, from the rise of agentic and AI-powered analytics to the growing importance of trust and collaboration.
Over the past year, we spoke with data practitioners and customers about their experiences using current BI tools: where they fall short, why confidence in AI-generated insights remains low, and what data teams can do to close the gap between promise and practice.
Together, these trends point to a future where analytics moves beyond static dashboards toward interpretable AI, collaborative workflows, and insights that meet people where they work.
Agentic and AI-powered analytics
AI is fundamentally transforming how businesses collect, analyze, and understand their impact through data. As BI and analytics platforms introduce more advanced agentic analytics capabilities, data teams and stakeholders are starting to move beyond asking AI about what happened in the past. Instead, they are beginning to rely on AI to suggest actions they should take next based on the data.
Even so, there’s a gap between the potential value of agentic AI and the real-world constraints that limit AI adoption. To better understand what’s holding AI adoption back, we surveyed data practitioners in late 2025. When asked how confident they were in the accuracy and reliability of AI-generated insights in their current BI tools, only 10% said they were somewhat or very confident. Everyone else reported feeling neutral to very unconfident.
This highlights a clear disconnect between the value AI tools promise and the trust users place in their outputs.
Their skepticism around AI for data analysis isn’t unwarranted: AI unavoidably hallucinates, misinterprets requests, and makes mistakes. The problem is that when AI works as a black box, users can’t see how an answer was produced, which makes it hard to verify whether a response is correct or appropriate for the situation. That’s why, at Observable, we are taking a transparent and human-centric approach to AI. Observable AI is designed to “show its work” by producing interpretable and editable queries and responses that keep analysts in the driver’s seat.
The growing importance of trust
Confidence in analysis can quickly erode when there are questions about data quality or how raw data was transformed into a chart, dashboard, or report.
Since mistrust is a major barrier to AI adoption for data analysis, improving the accuracy and verifiability of AI’s work will be increasingly important. Verifiability needs to be more thoroughly baked directly into AI tools and workflows. For example, when answering a data question, AI should return a query with clear provenance that a user can inspect. Giving users the ability to audit AI’s work is essential for driving adoption and closing today’s trust gap.
Another area where teams can increase trust in output is by making it easier to track data transformations — whether performed manually or by AI-generated queries. Making data visible in charts and tables throughout the analysis process can reveal new questions, anomalies, mistakes, and misunderstandings earlier. It also makes results easier to interpret across a wider audience.
In Observable Canvases, we keep data visible by default with visual summaries that preview the data at each step, and show concise summary charts for each variable. These visualizations aren’t just decorative; they are interactive interfaces for quick data exploration and deeper understanding.
Accessible, interactive data exploration fosters trust because users have the agency to dig into the data on their own. When a user can independently click on a bar or brush over a cluster to investigate different slices of the data, ask follow-up questions, or even have AI help explain what they’re seeing in context, they build confidence in their understanding and shared results. There is an important role for interactive data visualization, whether created by AI or a human, in increasing trust and transparency in analyses.
From self-serve analytics to collaborative analytics
Stakeholders increasingly want to explore data on their own. In part, this is because data teams are often overwhelmed by inbound requests, creating slow response times. More importantly, data can’t become part of everyday decision-making if every new question requires a handoff.
This shift toward self-serve analytics is a positive one, but it comes with tradeoffs. Many self-serve analytics tools limit deeper exploration or advanced, custom chart creation, resulting in shallow insights. Even worse, self-serve tools can unintentionally reinforce silos by creating the perception that stakeholders don’t need to interface with data teams at all.
Collaborative analytics, on the other hand, seeks to create more points of connection between data teams and their stakeholders by involving stakeholders throughout a project, instead of just tapping them for a random question or copy edits on a final report. This approach brings valuable perspectives and expertise to the entirety of the data analysis workflow from ideation to final data product. As Observable’s co-founder and CEO Mike Bostock put it, “Isolating roles is counterproductive; we should instead bring people together more efficiently.”
With the right tools and processes, data teams that embrace collaborative analytics create guardrails for stakeholders to empower their exploration.
Breaking charts out of the BI dashboard
Charts are already ubiquitous in consumer technology, and their presence in business software will only continue to grow.
Embedded analytics is expanding the domain of charts and insights beyond traditional BI dashboards, and into the tools people already use — from their company portal to messaging apps like Teams or Slack. These use cases require responsive, lightweight charts that work well in small spaces. An accelerator for this shift will be agentic AI. As chatbots and AI assistants become embedded in everyday workflows, they’ll proactively surface relevant charts and insights at the moment of need.
Bringing the chart to the user keeps insights visible and timely, prevents dashboard rot, and accelerates data-driven decision making. Most importantly, it allows employees to engage with data without breaking their flow or switching tools.
What this means for data leaders
Underlying each of these trends is an important shift in the role of data. The success of analytics initiatives isn’t determined by how many dashboards are created or how many views those dashboards received. Rather, success is evidenced by the role that data plays within the daily activities and decisions of employees. It is about democratizing data while maintaining consistency and guardrails, giving employees the tools they need to increase their own adoption of data in their respective roles, and embracing tools and practices that build trust in AI- or human-generated insights.