AI Studio · Coming in Gluesync 2.3

Build AI agents on the data plane you already trust.

Bring the model. Gluesync provides the tools, permissions, and operational context. Ask about pipelines, inspect the estate, or query data without creating a privileged AI account.

AI Studio · Product experience Governed session
You

Which unpaid invoices are more than 60 days overdue?

Agent

I’ll query the ERP source with your Query Studio permissions.

Tool calls
list_sql_capable_agents()execute_sql(…)
Agent

128 invoices found. The three highest balances total €89,830.

You

Why is orders-to-analytics falling behind?

Agent

I’ll inspect the pipeline, its entities, and source metrics.

Tool calls
get_pipeline_status(…)get_agent_metrics(…)
Agent

The source agent is delayed. 42 of 45 entities are still syncing.

You

Which connection does the billing pipeline use?

Agent

I’ll follow the allowed Core Hub tools and show every call.

Tool calls
list_connections()get_connection_usages(…)
Agent

Billing uses ERP-PROD-SQL. No credentials were exposed.

The control plane becomes the AI control plane

Useful enough for operations. Bounded enough for enterprise data.

01

Your model. Your keys.

Connect OpenAI, Anthropic, Azure OpenAI, or Ollama. Credentials are write-only in the Core Hub provider vault and never returned by the API.

02

A toolset for every agent

Choose from the Core Hub MCP catalog and saved Query Studio tools. An allow-list limits what each agent can see and call.

Explore →
03

Your permissions stay in force

Every invocation forwards the caller’s identity through existing Core Hub authorization. SQL follows Query Studio controls and is read-only by default.

Explore →
04

See how every answer was built

Follow tool names, outcomes, and sanitized context turn by turn. Traces are useful for review without exposing provider secrets or raw result rows.

Your infrastructure · your models · your permissions

See AI Studio before Gluesync 2.3 ships.