No Jitter Brief:
77% of data leaders said that 20% or less of their data is contextualized, according to a Teradata/Wakefield report. Because of this, agentic AI is operationalized. 40% of organizations have developing agentic AI maturity, meaning they have solid AI models, but lack context and unification for data.
The report says that a lack of data infrastructure results in paused AI pilot projects because organizations don’t yet have the foundation required to operationalize agentic AI.
“Context fragmentation is what makes the leap from personal AI to organizational AI so difficult. Building an agent that works for one person is relatively straightforward. Building one that works across an organization requires significantly more effort because getting context right means knowing which data, assembled in what order, on behalf of which process, needs to reach an agent at the moment of a decision,” per Teradata/Wakefield.
No Jitter Insight:
Teradata describes four levels to agentic AI maturity: experimenting, developing, building and operationalizing. A majority of respondents (40%) were in the development stage, where they had AI models but no context of data unification. In this stage, data is siloed, meaning teams define, label and house their data separately rather than sharing it within the organization. 28% of respondents were still in the experimenting stage, where they had localized AI pilots and were mapping data strategies.
The building stage is where organizations begin to see governance and automated workflows. But the data foundation here is still localized, so organizations aren’t working with the same standardization and data isn’t unified. 25% of organizations were in the building stage.
The final stage, operationalization, is where many organizations want to be. Operationalizing means they have unified, governed data with multi-step workflows in place and actively developing context and lineage for data. Only 7% of respondents considered their organization to be at this stage.
Contextualized data is important for AI to understand what’s going on within the organization. In conversation with No Jitter, Dialpad CEO Craig Walker said, “Take agentic out of the equation and everything is handled by humans: They understand all the customer’s context, they have the full conversation, they solve the problem, they work through it. If you’re going to remove part of that human for that human equation, the agentic piece has to be able to have those same conversations.”
However, Josh Fecteau, Chief Data and AI Officer & Chief Information Officer, Teradata, says that contextualizing the entire data estate is the wrong goal and that it’s better to focus on the highest-value data first. “If most of the data is unusable, the answer isn’t to fix all of it at once. It’s to be ruthlessly selective about which 20-50% you start with.”