Agentic AI is moving quickly from experiments to real work. In 2026, it shows up inside the workflows that drive outcomes: decisions, operations, and accountability.
In the season 7 premiere of the Data Chief podcast, host Cindi Howson sat down with three leaders who work at the intersection of AI ambition and enterprise execution: Paul Baier (GAI Insights), Jennifer Belissent (Snowflake), and Rory Blundell (Gravitee). Together, they outline what’s changing in 2026 and the practical steps organizations can take now to keep pace.
If you’re a business leader funding investments, a data leader building the foundation, or a product leader navigating integration and time to market, we’re offering you a clear 2026 roadmap.
Quick ways to go deeper:
Podcast: The Data Chief season premiere with Paul Baier (GAI Insights), Jennifer Belissent (Snowflake), and Rory Blundell (Gravitee).
Ebook: ThoughtSpot’s 2026 trends guide with the broader market view and supporting research.
On-demand webinar: A separate panel discussion hosted by Cindi Howson and Jane Smith with , and Juan Sequeda (ServiceNow).
The Signal: Production Teams Are Pulling Ahead
One of the hardest hitting realities from this episode is the growing gap between “we’re evaluating” and “we’re operational.”
Paul Baier shared a benchmark from his enterprise-focused customer base: 25% of mid-size to large companies have something agentic in production, versus 8% reported by Gartner based on its broader customer base.
The point is not which number is “right.” The point is what those differences represent: teams in production are learning faster, improving faster, and building internal playbooks competitors can’t replicate overnight.
“So there really is a benefit to being first, and we continue to see that leaders are getting further ahead. There is no fast catch up.”
Paul Baier, CEO and co-founder, GAI Insights
Baier also emphasized that once a team learns to deliver agentic AI in one area, they can apply that capability to the next initiative and capture 80 to 90% of the benefits without starting from zero. That compounding effect is a major driver of early advantage.
Proof Point: Blue Cross Blue Shield of Michigan
Baier highlighted Blue Cross Blue Shield of Michigan (a $35B organization) as an example of what “ahead” can look like: 115 agents in production under CIO Bill Hammrich, already delivering material impact.
The takeaway for 2026 is straightforward: teams that build real production experience now are likely to widen their lead.
Data Readiness is the Real AI Strategy (And the Bar Is Rising)
Jennifer Belissent’s message is a practical one: the fundamentals haven’t changed, but the stakes have.
As AI moves from “answering questions” to supporting operations, organizations become more dependent on fresh, high-quality, secure data, not less. As Belissent puts it, “there’s no AI strategy without a data strategy.”
“We are all going to need to be AI managers.”
Jennifer Belissent, Principal Data Strategist, Snowflake
In 2026, “AI-ready data” needs to hold up under pressure:
Data fresh enough for decision-making, not stale reporting cycles
Security and governance that enable reuse safely, not repeated one-off builds
Relational and sem-structured data
This is also where product and engineering leaders become central. Agentic systems don’t just need clean data - they need reliable interfaces to act.
The Shift Many Teams Miss: Governance Can Speed Adoption
Belissent challenged a common assumption: AI governance and regulation can accelerate adoption because transparency builds trust, and trust enables reuse.
Here’s the mechanism she described:
When teams can see what exists (model registries, data catalogs, data products), duplication drops.
When reuse becomes normal, value grows with minimal incremental effort.
When transparency is a requirement internally, it can drive stronger collaboration across teams.
She shared an example: a retailer built a model for routing store pickers, then adapted it for logistics delivery. Faster time-to-value, less reinvention.
This is a core move for 2026: treat AI governance as a scaling system, not a constraint.
MCP Servers, Agent Protocols, and the New Integration Reality
The story isn’t “APIs are dead.” It’s that integration is expanding to support new users: agents. Rory Blundell described the last year as the first real wave of agent-to-tool integration, with 2026 focused on scaling those patterns across the enterprise.
“2025 has been the year of integrating agents.”
Rory Blundell, CEO, Gravitee
His key clarification is important for product and platform teams: APIs remain essential, but agent protocols are additive. Traditional APIs still support human-led interactions, while features like Model Context Protocol Servers (MCP servers) help agents and LLMs understand and use tools more effectively.
He shared a simple example that illustrates the maturity gap:
He built an agent that could recommend a restaurant
It failed when it came time to book the table and pay a deposit
Not because the model lacked capability, but because the ecosystem for action is still maturing
That’s the 2026 challenge in one moment: insight is often easier than action.
The Discipline That Decides Winners: Agent Management
Blundell offered the term that ties it together: agent management, essentially “API management for the agent era.”
He also described a maturity curve many enterprises are navigating:
Pilots and proofs of concept
First agents in production, often siloed
Interconnected agents that coordinate and share context
Governed multi-agent ecosystems built for trust, security, and control
Most organizations are somewhere between stages 1 and 2 because they do not yet have the security, governance, and controls to confidently move from “informational” agents to “action-oriented” systems.
His prediction is clear: in 2026, the companies that adopt robust security frameworks will accelerate fastest.
Want to leverage the next layer of intelligence? Get the Leader’s Guide to MCP.
Leadership Alignment: Move from Headlines to Hands-On
Baier’s most actionable guidance was about leadership behavior.
He argued executives should not be making or influencing GenAI investment decisions without real hands-on exposure:
At least a 90-minute leadership session where half the time is hands-on keyboard
And a higher bar: 30 hours of personal tool usage before weighing in on strategy and spend
Belissent reinforced the parallel: as assistants become agents, everyone becomes a manager. That means job descriptions, onboarding, performance evaluation, and accountability, similar to how teams manage people, but adapted for AI.
This is what “executive AI readiness” looks like in practice in 2026: fewer abstractions, more real experience.
Where ThoughtSpot Helps Teams Operationalize Agentic AI
A theme that keeps appearing is that agentic analytics only works when people can ask questions, trust the answers, and apply insights in context.
If you’re evaluating production AI systems in analytics, the core blockers the guests outlined stay consistent:
Data freshness and quality
Trust and transparency
Governance that enables reuse
Integration patterns that support action safely
Leadership literacy and accountability
Want to see what this looks like in practice? Start a free trial of ThoughtSpot.
Final Takeaway: 2026 Is the Year You Operationalize
Agentic AI in analytics will not arrive as a single “launch moment.” It will show up through operational decisions, made across teams.
The organizations best positioned in 2026 will:
Treat AI as a capability, not just an IT initiative
Invest in AI-ready data and trusted governance
Build agent management and security early
Create leadership literacy through hands-on practice
If you want the tightest “from trends to execution” path:
Start with the Data Chief episode: Cindi Howson with Paul Baier (GAI Insights), Jennifer Belissent (Snowflake), and Rory Blundell (Gravitee).
Then grab the ebook: ThoughtSpot’s 2026 trend map, led by Cindi Howson and Jane Smith.
And watch the webinar (on-demand): Cindi and Jane in conversation with Juan Sequeda (ServiceNow).
