5 predictions about agentic AI and analytics in 2026

When it comes to AI, there’s a question floating around every organization: how are other companies planning for agentic systems in 2026? 

Gartner set out to answer questions like these in a new predictions report. The report titled Predicts 2026: AI Agents, MCP and Governance are Transforming Analytics, surveyed organizations on their strategic planning for agentic AI.

Peter Corless, Principal Product Marketing Manager at Redpanda, compared Gartner’s findings to his own industry observations in a recent Tech Talk, and shared his predictions for the AI and analytics landscape in the coming months.

In this post, we cover what Gartner’s data showed as well as what Peter believes is on the horizon for analytics and AI. You can also watch Peter’s full Tech Talk on the Top AI Trends Shaping Analytics Through 2026.

How are companies planning for AI and analytics?

Gartner’s report reaffirms what we’ve already hypothesized (and our reason behind developing the Redpanda Agentic Data Plane): while agentic AI creates new opportunities to work smarter, it also comes with inherent risks and complex governance challenges. 

By 2028, Gartner predicts:

• 60% of self-service analytics users will use general-purpose LLMs for ad hoc and exploratory analysis while production-grade reporting will remain in traditional analytics and business intelligence platforms
• 60% of agentic analytics projects relying solely on MCP will fail due to the lack of a consistent semantic layer
• 25% of ungoverned decisions using LLMs will cause financial or reputational loss due to human biases, insufficient critical thinking, and AI sycophancy

{{featured-resource}}

Interestingly, when Peter surveyed his Tech Talk audience on the same topics, the responses differed from some of Gartner’s predictions. Watch the full talk to see those survey results.

What are the top predictions about agentic AI and analytics in the coming months?

Now we know what Gartner says, but what AI trends and challenges does Peter think organizations should be paying attention to?

Leaning on the many conversations he’s had at recent meetups and events, Peter compiled his top five predictions about agentic systems (particularly as they relate to analytics).  

1. Organizations that succeed at AI will share stakeholder duties across the leadership team

Historically, many transformative IT technologies have been relegated to a single department head—often the Chief Information Officer (CIO) or Chief Technology Officer (CTO). But for an organization to successfully implement agentic systems, Peter points out the need for across-the-board collaboration from the executive team.

While different C-suite leaders focus on different pieces of the AI puzzle, everyone needs to share responsibility for project success or risk failure.

2. Teams will need to learn new skills and work together in new ways to operationalize agentic systems 

Just as the executive team will need to come together, so too will the different roles involved in these various domains. All of these roles will need to learn new skills and best practices while collaborating across teams to build effective agentic systems. 

3. Data analysts will be one of the top roles in need of retraining to meet AI demands

Speaking of roles in need of new skills, Peter predicts data analysts will undergo one of the most extreme shifts around responsibilities and skillsets. AI is likely to reduce a data analyst’s manual workload for data cleaning, transformation, and basic visualization from 60–70% down to 20–30%. 

As a result, data analysts will shift from “query writing” to “AI orchestrating”.

4. Vibe coding will hit a wall without executive backing and guardrails

Vibe coding has gained popularity because it allows virtually anyone to use AI to develop applications (without the knowledge or time required to code manually). But enterprise-level vibe coding poses significant operational and security risks.

According to Peter, the immense risks involved in moving vibe coding to full-scale, enterprise-grade production will halt many such projects. 

Check out our post on O’Reilly Radar about the risks of AI coding without a plan:  Don’t Automate Your Moat: Matching AI Autonomy to Risk and Competitive Stakes.

5. Governance will become the leading topic in enterprise-level conversations about AI

Governance seems to be the buzzword of the year when it comes to agentic AI (and for good reason). Peter expects this will remain a hot topic as organizations look to solve the inherent challenges of deploying agentic systems.

Peter also notes the emerging AI frameworks and standards that many companies will start to adopt. Here are two to give you an idea:

• ISO 42001: An AI Management Systems (AIMS) framework that’s very similar to the ISO 27001 standard for information security management systems.
• AIUC-1: The equivalent of SOC 2 for AI agent safety, security, and reliability; integrates the theories of ISO 42001 into an actual rubric outlining business compliance. 

See the full list in Peter’s full Tech Talk! 

How can companies address AI governance challenges?

With these predictions in mind, Peter believes more organizations will start looking for a unified way to manage agentic systems. Companies need a framework to help address the disconnect between agents and data, and to ensure trust, explainability, and context. 

• Trust (governance): The ability to document responsibilities around security, safety, legality, and fiscality. 

• Explainability allows companies to observe systems for model performance, accuracy, and whether agent output is sensible.

• Context: The ability to connect to registries for schema, agents, tools, and Model Context Protocol (MCP) servers.

The Redpanda Agentic Data Plane was designed to help you operate and scale agentic systems while delivering governance and control in a single platform. ADP is built on top of the Redpanda streaming infrastructure that many know and love, and it can integrate with 300+ data sources via Redpanda Connect.

Watch the full tech talk

Peter’s predictions and the interesting differences between Gartner’s insights and the live audience’s thoughts are definitely worth a watch. This post was just an overview, so grab a brew and watch Peter’s full Tech Talk to learn: 

• The top insights from Gartner’s 2026 report on AI agents and governance
• How those insights compare to the live audience’s responses
• The specific skills data analysts will need to master in the agentic age
• The emerging AI frameworks and standards companies will start adopting
• How Redpanda’s Agentic Data Plane makes it simple to run and govern agentic AI

If you’re ready to go further and learn how ADP can help your organization safely scale agentic systems, get in touch with our team!