CData Connect AI

AI tools are becoming part of everyday business workflows, but they can also create new data security concerns. Employees may accidentally share sensitive information such as customer details, credentials, internal documents, or confidential logs with AI systems.One challenge is knowing what data is being sent to an AI service and whether sensitive information has been removed before processing. Access controls, data classification, monitoring, and redaction can help reduce unnecessary exposure.For teams using OpenSearch or other security platforms, I’m interested in how others are approaching this problem. Are you monitoring AI-related data flows, restricting access to sensitive fields, or using automated detection and redaction before data reaches an AI system?Question:What approach are you using to prevent sensitive data from being exposed through AI tools?

We should always remember that we are working on Scaling Laws & Emergent Abilities.... What are scaling laws in large language models, and how do model size, training data, and computational resources affect performance? Why might certain capabilities appear suddenly as models become larger? RegardsSoumitra Dutta

I am encountering this error since yesterday, JournalEntries seems OK but the JournalEntryLineItems returns this error. 

The CData Connect AI MCP integration lets you build an AI document agent that connects directly to Google Drive, reading actual document contents, reasoning across multiple files, and returning structured answers with source citations, without manual file searches or reformatting.In this article, you'll learn how to:Set up a Google Drive connection in CData Connect AI Enable Execute permissions for document content retrieval Convert .docx files to native Google Docs format for full content access Configure a Claude project with a system prompt to define agent behavior Validate the connection and run document queriesTo learn more, read:Connecting Google Drive to an AI Document Agent with CData Connect AI

Hey CData Community!CData Connect AI now supports HIPAA-regulated workloads involving PHI, we operate as a business associate for eligible use cases, and a BAA is available.The Data Privacy and Security article has been recently updated to reflect this: https://www.cdata.com/kb/entries/data-privacy-security.rstFull details at https://www.cdata.com/security/hipaa/. Need a BAA? Talk to your sales rep.

Upon logging in to my Connect AI account, I encountered:Error 500Internal Server ErrorAnd my Power BI Cdata connector also can’t connect to Quickbooks.Is this only on my account?

We published a step-by-step replication guide and benchmark kit that shows how Connect AI features like Custom Tools, Workspaces, Derived Views, and Caching dramatically reduce schema discovery overhead and AI costs. The guide includes:Full benchmark methodology Setup instructions for Connect AI GitHub benchmark harness Reproducible results and charts Read the full guide: https://www.cdata.com/kb/articles/token-benchmark-replication-guide.rst

Hi everyone, I’m Soumitra Dutta, an Oxford-based entrepreneur and photographer. Glad to be here! I’m running into some syntax issues while working with CData drivers and was hoping to get your advice.Some of my queries are either throwing errors or not behaving as expected, especially when using things like JOINs, filters, or ORDER BY clauses.Would really appreciate any suggestions or examples from your experience.RegardsSoumitra Dutta 

Just discovered CData Connect Spreadsheets when searching for a way to get Google Calendar information into Excel. Looks very good.However, when test-driving the software, I run into an issue.I have a primary calendar under my Google Workspace account [removed by moderator] . In addition, I have created multiple secondry calendars for various purposes.I've successfully connected the addin to my account, including successful authentication. When I start the Import process, I can select the configured connection, and the Table list populates with all available tables, including all calendars. When I select one of the secondary calendars, such as 'Placeholders' or 'Flags', the column list is loaded, the generated query is created, and the Execute/Reset buttons become available (see 2nd screenshot). When I hit Execute, the Excel list is populated with data from the selected calendar. All as expected.But when I select my primary calendar [removed by moderator] , none of that happens. The

Hi Everyone, I am Dheeraj Sudan , a UK-based software developer. I’m running into OAuth token authentication errors when trying to connect GA4 and a few APIs through CData. I’ve double-checked my client ID, secret, and redirect URIs, but the connection still fails. Has anyone experienced this before? Any tips or steps to troubleshoot and fix these authentication issues would be greatly appreciated! RegardsDheeraj Sudan and Meenu Hinduja 

You can now build custom MCP tools directly in Connect AI for agentic workflows.With Custom Tools, each operation is pre-defined with scoped inputs, validated logic, and structured output. This means your AI takes action on your terms — consistently, across every team that uses it. 

Connect Supabase to CData Connect AI to give AI tools secure, live access to your data—no SQL, no pipelines.What you can do:Query Supabase using natural language (ChatGPT, Claude, Copilot, MCP agents) Write back to your database without direct SQL access Query across 350+ connected data sources in a single request Maintain governed, real-time access to your dataRead the full guide: https://www.cdata.com/kb/articles/cloud-supabase.rst

ChatGPT encompasses multiple OpenAI model families including GPT-5, GPT-4.1, and o-series reasoning models designed to support natural language queries and tool-based data access. Integrated with CData Connect AI, ChatGPT enables secure, real-time enterprise data connectivity while maintaining compliance and governance standards.In this article, you'll learn how to:Understand ChatGPT's model families and enterprise capabilities Leverage native tool calling for seamless data access workflows Integrate ChatGPT with CData Connect AI for secure connectivity Navigate authentication, rate limits, and compliance requirements Implement multi-step analytics workflows with automated SQL generationTo learn more, read:Know Your LLM (Series): ChatGPT

Hi, CData team,  we found that the following query text stops works from 2025/1/26.SELECT [[char\]] FROM [SQL_KrewTestDB].[krewtestdb].[bug2149] LIMIT 500My table as following, the data source is My SQL. You can see that the column name is [char], it has the special char “[”, “]”, before it works well, but from 2025/1/26, it stop working. 

Choosing the right LLM for your enterprise data workflows just got easier.We've published a new Know Your LLMs series in our Knowledge Base—comprehensive technical evaluations of the leading AI models for integration with CData Connect AI. Each article covers architecture, API specifications, function calling capabilities, security posture, and deployment guidance.The series includes:Mistral AI — EU-hosted, open-weight models with native Le Chat MCP support ChatGPT (OpenAI) — GPT-4o and the industry's largest ecosystem Claude (Anthropic) — Extended context windows and strong reasoning Gemini (Google) — Deep Google Workspace integration Grok (xAI) — Real-time data and unique training approachEach evaluation includes verified benchmark data, integration patterns, compliance frameworks, and model selection guidance to help you match the right LLM to your use case.Have questions or want to share your experience with a specific model? Drop a comment below!

AI-driven analytics can be slow and costly when LLMs repeatedly rebuild schemas and relationships across multiple data sources. Derived Views with CData Connect AI eliminate that overhead, reducing token usage while making multi-source analytics faster and more reliable at scale.Read the full article here: https://www.cdata.com/kb/articles/connect-ai-derived-views-performance.rst

Hey CData Community!Tired of Claude guessing your schema wrong?Give it a brain upgrade with Claude Projects + CData Connect AI.We put together a quick guide on teaching Claude your database layout so it stops hallucinating columns and starts writing clean, accurate queries. It’s simple, it’s powerful, and honestly… it feels like giving Claude glasses. 🤓Check it out here: https://www.cdata.com/kb/articles/claude-projects.rst

Want to build intelligent agents that can query live enterprise data — no replication, no ETL, just governed, real-time access?CData Connect AI and IBM watsonx Orchestrate make it possible in minutes.What you can do Connect watsonx Orchestrate directly to 350+ data sources through CData’s MCP server. Use secure, governed SQL access — your data never leaves its source. Build agents that query, summarize, and act on real-time data. How it works Add your data connection in CData Connect AI. Link it in IBM watsonx Orchestrate using a secure Key-Value connection. Import CData MCP toolkits via the ADK CLI and start building workflows.   Try the tutorial for your platform 🔗 Salesforce Tutorial 🔗 Snowflake Tutorial 🔗 Jira Tutorial 🔗 SAP Tutorial Get startedStart building your own AI agents today with a free Connect AI trial here.

Learn how to bring live, conversational data insights directly into Microsoft Teams using Microsoft Copilot Studio and CData Connect AI. This guide walks through deploying Copilot Studio agents to Teams and connecting them securely to enterprise data via the Connect AI Remote MCP Server. With CData Connect AI, you can query and act on live data for smarter business interactions; no replication required.Read the full article here: https://www.cdata.com/kb/articles/cloud-copilot-teams.rst

We’re excited to announce our first-ever CData Connect AI Hackathon! 🎉 This is your chance to show off the coolest, most impactful AI use cases using CData Connect AI and MCP-enabled AI tools to connect live enterprise data to your agents, assistants, and copilots. Anyone can enter and win – Connect AI only takes minutes to set up.  Kickoff: September 24, 2025  Submission Deadline: October 31, 2025  How it works  Build your solution  Use CData Connect AI to connect at least one live enterprise data source.  Integrate with any MCP-enabled AI agent, assistant, or copilot.  Create something that’s innovative, practical, or just plain fun!  Submit your entry below  Post in this thread with:  Title of your solution  Description – what your solution does  Source systems connected  AI tools or agents used  Business impact – what problem it solves or opportunity it unlocks  (Optional) Demo – screenshots, videos, or links to show it

https://www.cdata.com/cloud/   Try it free, but it gives Error 500, please take a look.