Data lake

DuckDB

< Back to All Connectors
Connect your AI to DuckDB
Data engineers, finance analysts, and auditors keep years of exported ERP history and operational extracts as Parquet and CSV files in Azure Blob Storage, Amazon S3, or OneLake, but answering a question from those files usually means standing up a warehouse or writing SQL by hand. AI can query them directly, but only if it reads the files read-only, stays inside the storage root each request is granted, and never shares one tenant's credentials with another.

‍

Start a Trial
Get a Live Demo

Why it matters

Capabilities that matter most for DuckDB

01

Archived ERP data stays answerable

Read Dynamics GP history exported by the Popdock Data Lake Exporter as Parquet, with its lists, fields, relationships and companies intact. Retire the old server and keep the answers.
02

No warehouse to run

Popdock reads Parquet and CSV in place with an in-memory DuckDB engine. Drop a file in the container and it shows up as a list, with no loading, catalog or schema work.
03

Azure, S3 and OneLake in one connector

Point it at an Azure Blob container, an Amazon S3 or S3-compatible bucket, or a Microsoft Fabric OneLake lakehouse. AWS access can use an eOne-managed role, so you never hand over long-lived keys.
03

Read-only and sandboxed by design

The connector never writes to storage, and every query runs in a fresh, locked-down instance limited to your storage root. Credentials and data are never shared between tenants.
05

Governed AI over your files

Expose file lists as MCP tools so any AI client can filter, sort and summarize them, within the fields and companies Popdock lets each user see.
06

History next to live data

Join archived records with live data from Business Central, Salesforce or SQL Server in one Popdock list, so year-over-year answers span the old system and the new one.

Use cases

Example AI Use Cases for DuckDB

Filters

scenario

-

Years of exported ERP history and operational extracts sit as Parquet and CSV files in Azure Blob Storage, Amazon S3 or OneLake. Answering one question from them means finding the right file, loading it into a query tool and writing SQL by hand, so most questions wait for the one person who knows where the data lives.

What popdock provides

--

Ask about your archived tables and files in plain words, and PopdockAI reads the lists your admin exposed on the DuckDB connector to answer, with no SQL required. The files are only ever read, never changed, and the storage credentials stay in Popdock rather than in the AI client. Your admin decides which tables, fields and rows each user, team and AI client can see, with fixed filters the AI cannot remove, and answers come back as text, a table or an Excel file.

Business outcome

---

Questions about archived and exported data get answered in minutes, from the files themselves, without a warehouse or a SQL request queue.
See Full Use Case
No items found.

ready, go

Tools for DuckDB

Start with ready-made MCP tools for the DuckDB. Each one gives your AI client a single, clearly defined job: get invoices or journal entries, pull a file, or create and update records. Add a tool to your PopdockAI MCP server, choose the fields it returns and who can use it, and it's ready in Claude, ChatGPT or Microsoft Copilot. Nothing is exposed that you haven't switched on.

Tool Types

Name of tool
Description
No items found.

Ready to unlock your data?

‍

Start building tools that turn your business systems into AI superpowers

Join forward-thinking companies that are transforming their operations with intelligent, secure AI automation.

Try PopdockAI Free
See Documentation