Astrato's AI works inside your dashboards – generating contextual summaries, flagging points of interest, and answering natural language questions with live visualizations.
Every answer is grounded in your semantic layer, so AI speaks your business language and returns the numbers you can actually trust.

AI Insights generates a contextual summary of whatever data is on screen – trends, outliers, key drivers. Change a filter and the summary regenerates. Drill into a region and the narrative follows. Every insight reflects what the user is looking at right now, not what was true when someone last built a report.

Configure system prompts to tailor the output by role, industry, or business rules. A finance lead sees margin analysis. An ops manager sees fulfillment exceptions. Same dashboard, different level of detail.

Custom Report is a self-service sandbox where your users explore data on their own – slicing, filtering, and building visualizations from predefined measures and dimensions your team controls. Or they can ask a question in natural language and get a visualization back in seconds. One environment, two ways to explore.

BI Copilot handles the repetitive work your team does every build – generating chart titles, renaming database fields into business-friendly labels, and creating complex measures like YOY growth or customer retention from a plain-language description. Type what you need, review the output, and move on.


Every AI-generated answer – whether it's a summary, a visualisation, or a measure – draws from the business logic, metrics, and definitions your team has already built. AI doesn't interpret raw columns. It reads your semantic layer. That's why "revenue" means revenue, not a best guess.

Choose the model that fits your data policy. Astrato supports Snowflake Cortex, OpenAI, Azure OpenAI, Claude by Anthropic, and Google Gemini – or use the default provider included with your Astrato licence.


With Snowflake Cortex, queries execute inside your warehouse. Astrato sends the SQL, Cortex processes it, and only the text response comes back. Your data never leaves. With external providers, admins control exactly which AI features are enabled and what data each one can access – feature by feature, toggle by toggle.

No retraining required. The AI queries your Semantic Layer in real time, so when you add a new metric or rename a dimension, the AI picks up the change immediately. There's no model to retrain, no prompts to rewrite — the semantic layer is the AI's only source of truth.
No. When you use Snowflake Cortex or your own LLM, queries run inside your secure environment — only the generated text response comes back. Raw data never leaves your warehouse. You can also bring your own LLM provider and API key to keep everything under your control.
Yes. Astrato integrates with Snowflake Cortex (which supports Claude, Meta, Mistral, DeepSeek), Google Gemini, OpenAI, and bring-your-own LLMs. You pick the model that fits your governance, cost, and performance needs — we don't lock you into one AI vendor.
The Semantic Layer does the heavy lifting. Every metric, dimension, and relationship is defined once, so when a user asks “what's our best-performing product?”, the AI queries a governed model instead of guessing. Clear definitions plus warehouse-native execution means answers you can actually trust.
Yes. AI Insights are context-aware — they respect the filters, slicers, and selections active on the page. Ask a follow-up question and it picks up where the dashboard left off, instead of starting from scratch.