live wire
▸IBM makes watsonx Orchestrate AgentOps, custom LLM judging and Bedrock-agent discovery generally availableIBM▸SchemaGate 0.1.45 fixes broken Oracle ADB wallet connections and an OCI stack pinned 28 releases behindSchemaGate▸PDI’s Amazon Quick procurement agent grounds spend answers in vendor, category and contract contextAWS Business Intelligence Blog▸BigQuery’s ML.METRICS example returns 0.84 accuracy but 0.30 macro-F1 on the same 100-row classification queryGoogle Cloud BigQuery docs▸SchemaGate 0.1.44 auto-selects sentence embeddings, lifting bundled-schema retrieval from 90/98 to 93/98SchemaGate▸SchemaGate 0.1.43 adds read-only SQL execution with per-principal table checks—and documents unauthenticated client assertionsSchemaGate▸Databox adds reusable AI Analyst Skills with personal/company scope, auto-matching and marketplace installsDatabox▸Fabric previews an AI builder for data-agent instructions, source guidance and example queriesMicrosoft Fabric▸Databricks trains data-agent retriever to stop early or spend bounded extra search steps, reporting 5.8-second latencyDatabricks▸ThoughtSpot adds SpotterCode coding agent to its Visual Embed PlaygroundThoughtSpot▸LongMemEval-S audit: 67–73% of restore-fixable 80k-budget errors came from evicted evidence under three policiesarXiv▸SchemaGate 0.1.42 adds dimension-aware retrieval and fixes complex multi-table SQL promptsSchemaGate▸Snowflake agent toolsets can silently drop inherited tools when callers lack accessSnowflake Documentation▸Looker’s VS Code extension reaches GA with MCP-assisted LookML generation, editing and validationGoogle Cloud Looker release docs▸IBM makes watsonx Orchestrate AgentOps, custom LLM judging and Bedrock-agent discovery generally availableIBM▸SchemaGate 0.1.45 fixes broken Oracle ADB wallet connections and an OCI stack pinned 28 releases behindSchemaGate▸PDI’s Amazon Quick procurement agent grounds spend answers in vendor, category and contract contextAWS Business Intelligence Blog▸BigQuery’s ML.METRICS example returns 0.84 accuracy but 0.30 macro-F1 on the same 100-row classification queryGoogle Cloud BigQuery docs▸SchemaGate 0.1.44 auto-selects sentence embeddings, lifting bundled-schema retrieval from 90/98 to 93/98SchemaGate▸SchemaGate 0.1.43 adds read-only SQL execution with per-principal table checks—and documents unauthenticated client assertionsSchemaGate▸Databox adds reusable AI Analyst Skills with personal/company scope, auto-matching and marketplace installsDatabox▸Fabric previews an AI builder for data-agent instructions, source guidance and example queriesMicrosoft Fabric▸Databricks trains data-agent retriever to stop early or spend bounded extra search steps, reporting 5.8-second latencyDatabricks▸ThoughtSpot adds SpotterCode coding agent to its Visual Embed PlaygroundThoughtSpot▸LongMemEval-S audit: 67–73% of restore-fixable 80k-budget errors came from evicted evidence under three policiesarXiv▸SchemaGate 0.1.42 adds dimension-aware retrieval and fixes complex multi-table SQL promptsSchemaGate▸Snowflake agent toolsets can silently drop inherited tools when callers lack accessSnowflake Documentation▸Looker’s VS Code extension reaches GA with MCP-assisted LookML generation, editing and validationGoogle Cloud Looker release docs
nl2sql.ai
guideACCESS CONTROL

ThoughtSpot’s two ChatGPT Work plugins split data analysis from analytics administration

Spotter answers governed questions; SpotterCode can configure live instances. Enterprises should treat them as different risk classes, not one integration.

By The News Desk· Sep 12, 2026the quick take — two AI hosts go live when you do

ThoughtSpot has put two distinct plugins into the ChatGPT Work Plugin Directory, and the important deployment detail is not that both run in chat. It is that they have materially different powers. Spotter is the analytics-facing plugin; SpotterCode can build against and act on a live ThoughtSpot instance. That makes separate approval, identity and audit policies the safe default. ThoughtSpot

One plugin answers; the other changes the system

ThoughtSpot says Spotter lets users ask natural-language questions, investigate metric drivers, forecast trends and follow answers back through search logic to governed definitions. Responses are grounded in ThoughtSpot’s semantic and context layer, while row-, column- and attribute-level controls continue to apply. ThoughtSpot

SpotterCode has a broader operational role. The company says it can generate embedded Spotter and Liveboard components, configure the Visual Embed SDK and authentication, adjust CORS and security settings, and call REST APIs for tasks including tenant creation and user management. Those are administrative actions, not just analytical reads. ThoughtSpot

The practical implication is a two-tier permission model. A business user who needs governed answers should not automatically inherit the identity or scopes used by a developer building embedded analytics. Conversely, a build agent should not run under a shared service account that obscures which person requested a tenant, user or configuration change. ThoughtSpot says connectors are permissioned individually and AI-initiated actions appear in audit logs available through its API or a SIEM; administrators can use those controls to keep the tiers distinct. ThoughtSpot

Four checks before rollout

  1. Register Spotter and SpotterCode separately. Give Spotter read-oriented access to approved models and analytics objects. Reserve SpotterCode for a smaller developer or platform group with only the REST actions its workflow requires. ThoughtSpot
  2. Preserve per-user identity. ThoughtSpot says its existing identity provider, SSO, SAML or OIDC federation and per-user enforcement remain in the path. Test that the user represented in ChatGPT maps to the same ThoughtSpot permissions seen in the native product. ThoughtSpot
  3. Alert on administrative actions. Stream AI-initiated audit events to the SIEM and flag tenant creation, user changes, authentication edits and security-setting changes separately from ordinary questions. ThoughtSpot
  4. Verify the data-handling claim. ThoughtSpot says the integration copies no data out and does not persist prompts. Security teams should confirm that behavior for their configured connectors, retention settings and downstream ChatGPT Work controls before approving sensitive datasets. ThoughtSpot

The launch expands ChatGPT Work from a place to consume governed analysis into a surface that can also build and administer analytics applications. The useful boundary is therefore not “AI versus no AI.” It is read and reason versus configure and act—and the access model should make that boundary visible. ThoughtSpot

sources

Filed by The News Desk. Corrections: desk@nl2sql.ai · Our standards →

comments · 0

    Comments are moderated before they appear. Your email is used once to confirm it is you — never shown, never sold. Corrections and questions get an answer from the desk when we have one.

    more from the desk