Article
AI Hub is coming — not another chatbot on your sync tool
AI Hub is coming this month — not another chatbot on your sync tool. Gluesync 2.3 puts AI in the control plane across 59+ databases.
This is how we envisioned AI-powered data integration.
Bringing AI to 59+ databases — including legacy systems, no matter how old.
That is the line for Gluesync 2.3. Not a side chatbot bolted onto a sync tool. AI that sits in the control plane you already operate, under the permissions you already grant, against the agents and schemas you already trust.
Why 2.3 is about AI in the Hub
Most “AI for data” demos assume a clean warehouse and a shared service account. Real pipelines do not.
You have Oracle next to Postgres next to something that still speaks journal receivers. You have RBAC that actually matters. You have operators who will not accept a model that invents SQL and runs it because a prompt looked confident.
Gluesync 2.3 puts AI where those constraints already live: Core Hub.
What is coming
AI Studio
A place in Core Hub to work with agents that know your pipelines and can ask your data — with the model you bring.
BYO LLM in Hub settings (your keys, your providers). Tool access stays inside Hub RBAC and allow-lists. Traces stay visible. You shape the agent; you do not outsource the permission model.
Query Studio AI
Natural language in. Proposed SQL out. Same editor, same schema, same permissions.
The helper drafts. You review before anything runs. It is not an auto-execute path. Grounding is the open agent and the catalog you already use in Query Studio.
Core Hub MCP
Expose permission-aware Hub tools to approved AI clients over MCP. Same brain the rest of Gluesync AI leans on: pipelines, entities, metrics, governed SQL helpers — under the caller’s identity, not a shared god account.
If you already use Core Hub MCP today, 2.3 is the broader AI surface built on that foundation.
SQL-callable AI on Query Forge
For teams that live in JDBC clients, 2.3 extends the idea further: ask from SQL against a virtual catalog such as gluesync.ai, and get an answer path that still respects Hub RBAC and does not silently run proposed statements for you.
Same Forge connection style you already know (jdbc:gluesync://…).
AI on legacy databases
The point of “59+” is not a logo wall. It is that the AI path should reach the same agents Gluesync already syncs — including systems that predate cloud-native everything.
If the data still matters in production, the AI surface should be able to meet it there, under the same identity and safety gates.
What does not change
- Your infrastructure stays yours.
- Your models stay yours (BYO).
- Your permissions stay yours.
- Drafts and proposals are not a free pass past SafetyGate / RBAC.
We are not asking you to trust a black-box “data AI” that bypasses the Hub. We are putting AI inside the Hub.
What’s next
AI Hub is coming this month — sooner than you might think.
Want a sneak peek of what’s coming? Book a demo with us today, or request early access to the beta.
If you want the short version: Gluesync 2.3 is AI-powered data integration across the databases you already run — including the old ones — without giving up the control plane.