By AI agents deployed in enterprise environments inherit whatever context they are given. When that context is missing or wrong, the agent produces wrong answers and takes wrong actions. A misunderstood definition of “enterprise revenue” or “customers at risk” is not a model failure; it is a context failure, and no model upgrade fixes it.
In this interview on TFiR, Raviv Levi, Chief Product & Technology Officer at CData, breaks down why enterprise context is the prerequisite to accurate and safe agentic AI, and how CData Connect AI Gateway enforces policy controls down to the individual record.
Guest: Raviv Levi, Chief Product & Technology Officer at CData
Show: TFiR
Here is what every platform engineer and enterprise AI architect needs to know.
Technical Deep Dive
Q: Why do AI agents produce inaccurate answers in enterprise environments?
Raviv Levi, Chief Product & Technology Officer at CData, explains that agents have intelligence and data access but lack the enterprise-specific definitions and metadata that give those inputs meaning. Terms like “enterprise revenue” or “customers at risk” carry different definitions in every organization, and without that context embedded in the agent’s operating environment, answers are statistically unreliable. CData’s testing across comparable tools shows accuracy sitting at 65% without proper context, regardless of which underlying model is used.
“When I say enterprise revenue in my company’s context, what does that mean? Does that mean the revenue of the company as an enterprise? Does that mean the revenue that is coming from our enterprise customers?”
Raviv Levi, Chief Product and Technology Officer, CData
Q: What is enterprise context for AI agents and what does it include?
Enterprise context is the collection of metadata, business definitions, data location mappings, and logic that tells an agent what a question actually means inside a specific organization. Levi frames it as the prerequisite layer beneath any agentic action: before an agent can do anything useful, it must resolve ambiguous business terms into precise, agreed definitions. Context is not documentation; it is a structured, queryable layer that sits between the agent and the data sources.
“All that information and metadata is what comprises the context, the enterprise context that helps AI tools and agents to really get the answer that you need.”
Raviv Levi, Chief Product and Technology Officer, CData
Q: How much does enterprise context improve AI agent accuracy, and what is the cost impact?
Levi cites CData’s own benchmarking: accuracy jumps from 65% to over 98% simply by supplying correct enterprise context, and this improvement holds regardless of the model used. Because lower-cost models can reach 98% accuracy with proper context, the dependency on the most expensive frontier models disappears. Levi puts the potential cost reduction at more than 100x, since organizations no longer need to route every query through the highest-capability and highest-cost model to achieve reliable results.
“The cost reduction could be more than 100x because now you don’t necessarily need the most intelligent model out there to get a correct and accurate answer.”
Raviv Levi, Chief Product and Technology Officer, CData
Q: What does an AI agent need beyond accurate answers in order to take real action?
Levi identifies two requirements for meaningful agentic action: a correct answer derived from proper context, and broad access to the tools and data sources where that action must be executed. Accurate answers without action coverage produce insight with no execution path. Action coverage without accurate answers produces confident wrong actions. Both layers must be present simultaneously for agents to be operationally useful.
“In order for the action to be meaningful, you need two things. You need to know what you’re talking about. So that’s the accurate answer. And now you need to be able to access a variety of tools and data sources.”
Raviv Levi, Chief Product and Technology Officer, CData
Q: How does CData Connect AI Gateway connect AI agents to enterprise data sources?
CData brings hundreds of pre-built data source connectors, developed over more than a decade, into the Connect AI Gateway platform. The gateway exposes these connectors alongside natively supported MCP servers from various vendors, and also provides MCP compatibility for vendors that have not yet shipped native MCP offerings. This means agents can reach SaaS platforms, cloud data sources, and on-premises systems through a single governed access layer, without requiring custom integration work for each data source.
“Even if some vendors don’t have native MCP offerings, we would support that. So now your agents are able to research, get the right answers and take actions in the corresponding avenues of your choice.”
Raviv Levi, Chief Product and Technology Officer, CData
Q: How does CData Connect AI Gateway enforce governance and prevent agents from taking unauthorized actions?
CData deploys a policy engine on the gateway that allows administrators to predefine and enforce access policies before any agent request reaches a data source. The critical architectural distinction is that agents operate on the data plane, passing through the gateway, but have no access to the control plane where policies are defined and managed. Administrators on the customer side and at CData control which agents can access which data sources, tables, rows, and individual records, and under what conditions.
“Agents don’t have access to that control plane. The agents that are leveraging the Connect AI gateway are operating on a data plane, they’re going through the gateway, so they’re going through the pipe, but they don’t have control over the pipe.”
Raviv Levi, Chief Product and Technology Officer, CData
Q: What visibility do enterprises have into what AI agents are actually doing?
Every agent action passing through the CData Connect AI Gateway is logged and can be exported for audit and compliance purposes. Levi describes this as full visibility and observability across agent activity, covering which agent accessed which resource, when, and under what policy conditions. This logging layer is part of the gateway’s design rather than an add-on, making audit-grade observability a default behavior rather than an optional configuration.
“Everything also gets logged and exported as needed so we have full visibility and observability.”
Raviv Levi, Chief Product and Technology Officer, CData
Resources & Documentation
- CData Connect AI Gateway, enterprise data connectivity and AI agent governance platform with MCP server support and record-level policy enforcement
***
👇 Click to Read Full Raw Transcript
Swapnil Bhartiya: As we all know that companies are moving from AI standards that used to just answer questions to AI agents that actually take action on their own. What new challenges do they pose to enterprises that are running into, you know, when it comes to challenges related to data context or governance?
Raviv Levi: Yeah, so it’s, it’s all about the context. As, as you correctly mentioned before, when we’re thinking about the ability of an AI entity to help us in our business, they have the intelligence, they have access to data, but they need guidance and they need guardrails and they need to be explained. When we ask for this question, what do we actually mean when I say enterprise revenue in my company’s context, what does that mean? Does that mean the revenue of the company as an enterprise? Does that mean the revenue that is coming from our enterprise customers? How do we even define our enterprise customers when I’m asking about the customers that are at risk for this quarter? How do we define customers that are at risk When I’m looking for a specific piece of information, where is it stored? All that information and metadata is what comprises of the context, the enterprise context that helps AI tools and agents to really get the answer that you need. That is step one. Even without getting into agents that are taking action, you have to get the right answer. Interesting enough, in comparisons with other tools that I shall not name right now, we’ve seen that just by being able to ask the right questions and put the right context in front of AI tools, the accuracy of the answers that we’re getting is jumping from 65% to over 98%. And that’s regardless of models that you’re using. So the cost reduction could be more than 100x because now you don’t necessarily need the most intelligent model out there to get a correct and accurate answer. Now let’s go to the action piece. That is the obvious next step. But in order for the action to be meaningful, you need two things. You need to know what you’re talking about. So that’s the accurate answer. And now you need to be able to access a variety of tools and data sources. And this is really one of the benefits that CDATA has. Some of the work that has been done over the last decade, if not more than that, there’s been hundreds and hundreds of data sources and connectors that are now being made available through the platform, in addition to natively supported MCP servers that different vendors would offer. But more interestingly, even if some vendors don’t have native MCP offerings, we would support that. So now your agents are able to research, get the right answers and take actions in the corresponding avenues of your choice. And just as a quick anecdote and example, I have an agent that is doing exactly that. So my work agent is connected to all my SaaS and cloud data sources, obviously through Connect AI gets the right answers and is able to take action, whether it’s to set a meeting for me, update a JIRA ticket, update a Salesforce ticket, create slide decks and so on completely autonomously. And nine out of ten times I would say fully accurate.
Swapnil Bhartiya: How does Connect AI Gateway actually give organizations control over over what an agent can access and do? Write down to the individual record. Also considering that agents they can break all boundaries, they can do whatever they want to do as we have seen hugging face and such incidents in past as well. So how do you ensure that they do have that control?
Raviv Levi: We have a very sophisticated policy engine that we have deployed on the gateway, allowing our customers to predefine the policies and to enforce them. Now it’s important to say that agents don’t have access to that control plane, right? So the agents that are leveraging the Connect AI gateway are operating on a data plane, they’re going through the gateway, so they’re going through the pipe, but they don’t have control over the pipe. That control is reserved to administrators on the customer side, on the CDATA side that are able to define who has access to what, which data sources, which tables, which rows, which records, under which circumstances. And as mentioned before, everything also gets logged and exported as needed so we have full visibility and observability.
