Docs: Add capabilities page

docs/automation.mdx

@@ -1,5 +1,5 @@
 ---
-sidebar_position: 1.7
+sidebar_position: 1.8
 sidebar_label: Automation
 description:
   "Learn how to build automated workflows and processes with PromptQL for reliable, repeatable business tasks."

docs/capabilities.mdx

@@ -0,0 +1,117 @@
+---
+sidebar_position: 1.6
+sidebar_label: What can PromptQL do?
+description:
+  "Learn what PromptQL can do to help you make decisions quicker or automate tasks with relability and accuracy."
+keywords:
+  - promptql
+  - accurate
+  - reliable
+  - capabilities
+toc_max_heading_level: 4
+---
+
+import Thumbnail from "@site/src/components/Thumbnail";
+
+# What can PromptQL do?
+
+## Introduction
+
+PromptQL combines data analysis, AI capabilities, and automation tools into a conversational interface. It's designed to
+help you explore data, make decisions, and create reliable workflows—all through natural dialogue.
+
+## Core features
+
+### Agentic Semantic Metadata Layer
+
+PromptQL doesn't just execute queries—it builds and maintains a dynamic understanding of your data landscape. This
+"semantic metadata layer" grows more sophisticated with each interaction, learning your business terminology, data
+relationships, and analytical patterns.
+
+When you correct a definition or clarify a business rule, PromptQL remembers. It uses this knowledge to:
+
+- Translate natural language to precise queries
+- Apply consistent business rules across analyses
+- Suggest relevant connections between datasets
+- Self-correct based on feedback
+
+Think of it as an intelligent layer between you and your data that learns and adapts. It remembers that "active users"
+means something specific in your business, or that certain metrics should always be filtered in particular ways.
+
+:::info Learn more
+
+You can learn more about PromptQL's semantic understanding [here](/data-modeling/overview.mdx).
+
+:::
+
+### AI Primitives
+
+At its core, PromptQL works with data and AI functions, which we call **AI primitives**. It's helpful to think of these
+as tools in PromptQL's toolbox.
+
+For data, PromptQL can run complex SQL queries with built-in safety checks and optimization. It handles joins, and
+aggregations naturally, with automatic error handling and self-correction.
+
+The AI functions extend these capabilities beyond pure data analysis:
+
+| Primitive | Description                                  |
+| --------- | -------------------------------------------- |
+| Classify  | Sort text into dynamic categories.           |
+| Summarize | Create concise summaries of longer text.     |
+| Extract   | Pull structured data from unstructured text. |
+| Visualize | Create interactive data visualizations.      |
+
+By combining these primitives, PromptQL can tackle complex analytical tasks. For example, it might extract structured
+data from a batch of documents, classify the results, and visualize the patterns—all in a single workflow. Or it could
+summarize a large dataset, then extract key metrics from those summaries for deeper analysis.
+
+These combinations aren't just powerful—they're precise. Each primitive builds on the others, with error checking and
+validation at every step. The result is reliable, reproducible analysis that can be turned into automated workflows.
+
+:::info Learn more
+
+You can learn more about these primitives [here](/promptql-apis/execute-program-api.mdx#introduction).
+
+:::
+
+### Artifacts
+
+Think of artifacts as PromptQL's memory system. When you're exploring data or building workflows, artifacts help
+maintain context and enable reuse.
+
+| Type          | Purpose                                  |
+| ------------- | ---------------------------------------- |
+| Table         | Store and reference structured data      |
+| Text          | Preserve documents and long-form content |
+| Visualization | Interactive charts and graphs            |
+| Automation    | Reusable, parameterized workflows        |
+
+:::info Learn more
+
+You can learn more about artifacts [here](/promptql-playground/artifacts.mdx).
+
+:::
+
+## Use cases
+
+### Make decisions
+
+When it comes to decision-making, PromptQL acts as your analytical partner. It can explore data, detect patterns,
+visualize trends, and run comparisons. Each analysis can be saved for future reference, shared with your team, or used
+as part of your audit trail.
+
+The key is that PromptQL doesn't just show you numbers—it helps you understand them. Ask it to explain its methodology,
+break down complex metrics, or look at the same data from different angles. It's built to clarify, not just calculate.
+
+[Learn more](/decision-making.mdx) about making decisions with PromptQL.
+
+### Automate tasks
+
+Once you've found a valuable analysis pattern, PromptQL can turn it into a reliable automation. These aren't just saved
+queries—they're full workflows that can process new data, generate reports, monitor metrics, or alert on conditions.
+
+Each automation comes with built-in error handling and can be parameterized to handle different inputs. Chain them
+together for more complex operations, or schedule them to run regularly. The goal is to turn your one-off analyses into
+reliable, repeatable processes.
+
+[Learn more](/automation.mdx) about automating tasks with PromptQL.

docs/decision-making.mdx

@@ -1,5 +1,5 @@
 ---
-sidebar_position: 1.6
+sidebar_position: 1.7
 sidebar_label: Decision Making
 description: "Learn how you can use PrompQL for accurate AI in your decision-making processes."
 keywords:

docs/how-to-talk-to-promptql.mdx

@@ -85,8 +85,8 @@ The best way to get started is to ask something simple but real. Start with a me
 analysis you've done manually before. From there, PromptQL will learn your expectations, surface useful patterns, and
 help you turn those insights into reliable, repeatable workflows.
 
-We recommend picking up with [our overview of PromptQL's capabilities](#) but, if you're eager, jump in to one of these
-use cases:
+We recommend picking up with [our overview of PromptQL's capabilities](/capabilities.mdx) but, if you're eager, jump in
+to one of these use cases:
 
 - [Making decisions](/decision-making.mdx)
 - [Automating tasks](/automation.mdx)