CData Expands Connect AI Platform with New Agent Tooling and Enterprise-Grade Security to Power Production AI Deployments

CData Connect AI achieves 98.5% accuracy across 378 real-world prompts, outperforming competing MCP providers by 25+ percentage points

ORLANDO, Fla., March 9, 2026 /PRNewswire/ -- CData Software, the data layer for AI, today announced major enhancements to CData Connect AI at the Gartner Data & Analytics Summit (Booth #308). The updates extend CData's managed Model Context Protocol (MCP) platform with new capabilities across connectivity, context, and control, the three pillars required to move AI from experimentation to production.

Why AI Stalls Before Production

AI investment is accelerating. "Gartner®¹ says worldwide AI spending will total $2.5 trillion in 2026." But spending isn't translating into results. Most generative AI initiatives still stall before reaching production. The bottleneck isn't model capability, it's the data infrastructure underneath. Without live connectivity to business systems, semantic intelligence that gives data context to AI, and governance controls that enforce security at scale, AI initiatives fail to deliver business value.

CData's own State of AI Data Connectivity Report reinforces this reality. Only 6% of organizations are satisfied with their current data infrastructure for AI. More than half still rely on custom-built integrations that can't scale. And 71% of AI teams spend over a quarter of their implementation time on data integration alone, time spent wiring plumbing instead of building intelligence.

"AI agents are only as effective as the tools they can access and the data behind them, and only as safe as the controls governing both," said Amit Sharma, CEO and Founder of CData. "This release gives teams the ability to build use-case-specific agent tools with the right business context, deploy them securely, and enforce granular controls over which data agents can access, which actions they can take, and under what identity. That's what's been missing, not better models, but the connectivity, context, and control that make agents trustworthy enough to run in the enterprise."

Connect AI: Connectivity, Context, and Control in a Single Platform

CData Connect AI is purpose-built to address the data infrastructure gaps that prevent AI from reaching production. Today's enhancements extend the platform across all three pillars:

Connectivity: Connect Gateway and 350+ Data Sources

Connect AI provides live, read-write access to more than 350 business systems, without replication or data movement. The new Connect Gateway extends this reach to data sources behind the firewall, with support for SAP, SQL Server, and PostgreSQL, and more. The result: AI systems can operate against live data regardless of where it resides.

Context: Expanded Agent Tooling and Toolkits

AI agents need business-aware context to choose the right actions and avoid unnecessary MCP tool calls. But exposing too much context creates new risks: increased token usage, model confusion, and unintended access to sensitive data or operations. Connect AI addresses this challenge with a scoped MCP architecture that precisely controls what each agent can see and do. This release introduces three complementary tool types:

• Universal Tools provide a normalized set of operations that work consistently across all 350+ connected systems. Instead of exposing hundreds of system-specific tools, agents receive a compact, schema-aware interface ideal for data exploration, ad-hoc analysis, and multi-source reasoning — without tool surface bloat.
  
  
• Source Tools expose tightly defined operations specific to each system. These tools map directly to approved system actions, allowing IT teams to enforce predictable execution, transactional safety, and auditability for production workflows.
  
  
• Custom Tools allow organizations to define purpose-built operations tailored to specific workflows. These tools execute pre-optimized queries with explicit data access limits -- reducing token usage, improving performance, and eliminating unintended data exposure.

Workspaces define the data boundary for each agent by specifying exactly which datasets, schemas, or views are accessible. New Toolkits define the action boundary by determining which Universal, Source, or Custom Tools are available. Each Workspace and Toolkit combination can be deployed as a dedicated MCP server, ensuring that agents operate only within their intended scope; reducing context noise, strengthening governance, and delivering enterprise-grade control over agent behavior.

Control: SCIM and Custom OAuth Applications

Connect AI enforces per-user authentication with native source-system permissions applied dynamically at runtime, backed by full audit trails. New governance enhancements include SCIM 2.0 for automated identity lifecycle management and Custom OAuth Applications that enable organizations to use first-party credentials to meet internal security and compliance requirements. Every query is authenticated, authorized, and auditable.

The 25% Accuracy Gap: Why Architecture Matters

MCP is becoming the default interface between AI agents and business software. But how accurately do MCP providers actually return data? To find out, CData tested five MCP providers, representing the major architectural approaches in the market, across four sources (CRM, project management, data warehouse, and ERP) using 378 real-world prompts. Every response was scored against pre-established ground truth. No partial credit.

The results revealed a significant accuracy gap. CData Connect AI achieved 98.5% accuracy (67 of 68 correct responses). The other providers ranged from 65% to 75%—failing on one out of every three to four queries. The failures weren't random: they clustered around relative date logic, multi-filter queries, semantic interpretation of business terms, and write operations, exactly the kinds of tasks AI agents need to perform reliably every day.

For organizations moving beyond copilots toward autonomous agents that read, write, and act on live business data, this gap is decisive. At 75% accuracy, an AI agent fails one out of every four actions. And that inaccuracy compounds: 75% accuracy across a five-step workflow means less than 24% of processes complete successfully. A 75% accuracy rate becomes a 75% failure rate.

Most MCP providers translate natural language directly into API calls, which works for simple lookups but breaks down when queries require date math, multi-condition filtering, or platform-specific business logic. Connect AI uses a relational abstraction layer with semantic intelligence that understands entity relationships, business conventions, and workflow rules. That's why it maintained near-perfect accuracy across every platform tested, including ERP, where the vendor's own native MCP server failed completely.

View the full benchmarking methodology and results here: https://www.cdata.com/lp/ai-accuracy-whitepaper/

"AnySoft is an agentic coding tool that builds software that fits your business by ingesting, enriching, and unifying an organization's live business data—giving AI the context it needs to power everything from CRM to marketing automation," said AnySoft CEO Alex Noe. "That means connectivity to every system our customers run on isn't a nice-to-have, it's foundational. CData Connect AI Embed gives them production-grade access to 350+ data sources with the control and compliance built into the data layer."

"Connect AI has been a game changer for how we retrieve information from our enterprise systems using AI and the reason it works so well is how it manages context," said Paul Kantorovich, Manager of FP&A and AI Strategy, Foodtastic. "It gives Claude intermediary steps to scope a request before retrieving data, unlike alternatives that load full tables and overwhelm the context window."

Organizations deploying AI in production need an accuracy rate that prevents autonomous agents from creating more cleanup work than they save. Connect AI is built to clear that bar because connectivity, context, and control aren't just platform features. They're what makes accuracy at scale possible.

CData at Gartner Data & Analytics Summit

CData will be at the Gartner Data & Analytics Summit at Booth #308, where attendees can connect with the team and see the latest in universal data connectivity.

Speaking Session: AI Agents and the Future of Digital Work with Microsoft — CData Chief Product Officer Ken Yagen will take the stage alongside Microsoft Partner Director of Product Management James Oleinik on Wednesday, March 11 (11:15–11:45 AM EDT). The session will present a joint blueprint for moving from AI pilots to production-ready agentic AI, exploring how Copilot Studio and universal data connectivity can deliver the governed infrastructure enterprises need as Gartner predicts 40%+ of agentic AI projects will be canceled by 2027 without the right architecture in place.

Supporting Resources

• The 25% Accuracy Gap: MCP Provider Performance Across Enterprise Workloads — CData's benchmark of five MCP providers across 378 enterprise queries found a 25+ percentage point accuracy gap, with CData Connect AI achieving 98.5% accuracy compared to 65–75% for other providers. Download the whitepaper: https://www.cdata.com/lp/ai-accuracy-whitepaper/
  
  
• The State of AI Data Connectivity Report: 2026 Outlook — Based on research with 200+ data and AI leaders and insights from AI pioneers at Microsoft, AWS, and Google, CData's report found that only 6% of enterprises consider their data infrastructure fully ready for AI — establishing a direct link between data infrastructure maturity and AI success. Download the report: https://www.cdata.com/lp/ai-data-connectivity-report-2026/

¹ Gartner, Inc., "Gartner Says Worldwide AI Spending Will Total $2.5 Trillion in 2026," Gartner.com (Jan. 15, 2026), accessed Feb. 20, 2026, https://www.gartner.com/en/newsroom/press-releases/2026-1-15-gartner-says-worldwide-ai-spending-will-total-2-point-5-trillion-dollars-in-2026
GARTNER is a trademark of Gartner, Inc. and/or its affiliates.

About CData Software

CData is the data layer between AI and ROI—delivering the connectivity, context, and control that make enterprise AI more accurate. One platform for live access and data replication across 350+ sources, semantic intelligence that ensures context-aware responses, and built-in governance to control every AI-to-data interaction. Powering AI workloads for Anthropic, Databricks, Microsoft, Google, Palantir, and more than 10,000 customers worldwide. Learn more at www.cdata.com or contact [email protected].

SOURCE CData Software