Skip to content

Docs: Add decision making guide #162

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
---
sidebar_position: 51
sidebar_label: With Automation
sidebar_position: 1.6
sidebar_label: Automation
description:
"Learn how to build automated workflows and processes with PromptQL for reliable, repeatable business tasks."
keywords:
Expand All @@ -14,7 +14,7 @@ toc_max_heading_level: 4

import Thumbnail from "@site/src/components/Thumbnail";

# Building Automations with PromptQL
# Build Automations with PromptQL

## Introduction

Expand Down Expand Up @@ -140,7 +140,7 @@ https://promptql.ddn.hasura.app/playground/automations/v1/box_office_analyzer/ru

:::

## Best practices
## Best Practices

1. **Start simple** - Begin with a straightforward use case to learn the process
2. **Iterate and improve** - Refine your automation based on initial results
Expand Down
136 changes: 136 additions & 0 deletions docs/decision-making.mdx
Original file line number Diff line number Diff line change
@@ -0,0 +1,136 @@
---
sidebar_position: 1.5
sidebar_label: Decision Making
description: "Learn how you can use PrompQL for accurate AI in your decision-making processes."
keywords:
- promptql
- accurate
- reliable
- decision making
toc_max_heading_level: 4
---

import Thumbnail from "@site/src/components/Thumbnail";

# Make Decisions with PromptQL

## Introduction

**Decision making** with PromptQL supports deeper analysis and structured exploration across your data. You can ask
complex, layered questions and get responses that adapt to your systems and terminology. PromptQL helps you drill into
root causes, compare across categories, and evaluate tradeoffs—without being limited by context windows or informal
language.

This is useful for scenarios that require exploration and judgment, such as:

- Investigating anomalies
- Comparing performance across teams or regions
- Understanding contributing factors behind trends

## Guides

Below, we've split out a few different use cases as examples. You can run these against the
[public sandbox-quickstart project](https://promptql.console.hasura.io/public/sandbox-movie-studio/playground).

### Q&A

It would seem that asking questions about data is simple, but this is difficult because business terms often map to
multiple systems or concepts. For example, "performance" might refer to revenue, engagement, or critical ratings,
depending on who’s asking. Systems don’t always agree on how those values are calculated, either.

Take this example:

> _How did our PG-13 portfolio perform against R-rated titles during the streaming transition period?_

PromptQL solves this by mapping ambiguous terms to precise system definitions and building a plan that retrieves data
from the right sources in the correct form. The generated plan accounts for relevant time windows, content ratings, and
distribution channels, producing a structured answer tailored to your domain.

<Thumbnail src="/img/get-started/playground-dm-streaming-results.png" alt="Details of a deployed automation." />

**Use this when you want direct answers that reflect your business logic and definitions.**

### Interrogation

PromptQL allows you to interrogate your data by following up naturally, asking for more detail, and adjusting the scope
as you go. The system preserves your analysis trail and ensures consistency across steps.

> _Looking at our 2015–2020 release slate, what's the correlation between our talent investment strategy and audience
> retention metrics?_

PromptQL solves this by generating a multi-step plan that fetches relevant datasets, applies statistical methods, and
structures results in a way that's easy to pivot or extend. You can modify thresholds, change groupings, or backtrack to
explore a different angle—all without losing context.

<Thumbnail src="/img/get-started/playground-interrogation.png" alt="Details of a deployed automation." />

**Use this when a surface-level answer isn't enough and you need to go deeper with confidence.**

:::tip Edit the query plan

Most AI tools are a black box: you don't know what's happening under the hood, how answers were arrived upon, or what
data was used.

With PromptQL, every response is backed by a transparent query plan that you can inspect, modify, and re-run. This gives
you full control over the logic, data sources, and assumptions behind each result—so you can refine, extend, or validate
the analysis as needed.

:::

### Deep Research

You can perform deep research that explores multiple hypotheses, benchmarks external data, and evaluates internal
patterns across time or categories.

> _Can you analyze the ROI patterns of our genre-blending titles compared to pure-genre releases between 2010–2020?_

PromptQL is unique because it treats research as a process—not just a query. It generates a plan that defines discovery
phases, collects and segments relevant data, and evaluates each hypothesis systematically.

<Thumbnail src="/img/get-started/playground-deep-research.png" alt="Details of a deployed automation." />

**Use this when you're trying to answer open-ended questions that require context and exploration.**

### Cross-Source Intelligence

Since you can join any source using your semantic metadata layer, PromptQL can resolve data across structured,
semi-structured, and unstructured systems in a single plan.

> _What's the risk profile of working with first-time directors who came from our star talent pool?_

PromptQL builds a plan that pulls structured records (e.g., director metadata), aggregates historical performance
metrics, and layers in qualitative signals from reviews or production notes. Relationships that span systems—like
casting history, sentiment, and audience reception—are captured and evaluated together.

<Thumbnail src="/img/get-started/playground-cross-source.png" alt="Details of a deployed automation." />

**Use this when your answers require stitching together multiple systems and surfacing insights that aren't visible in
any single source.**

### Smart Visualizations

Visualizations make it easy to understand complex patterns or communicate findings across stakeholders. PromptQL
automatically selects appropriate formats—charts, tables, or graphs—based on the type and scale of your analysis.

<Thumbnail src="/img/get-started/playground-visualization-rosling.png" alt="Details of a deployed automation." />

**Use this when you want to share findings with others or spot trends across segments or time periods.**

## Best Practices

- **Start specific, then expand.** Narrow, well-defined questions help PromptQL build better initial plans. You can
always widen scope through follow-ups.
- **Use your own terms.** PromptQL is designed to understand your internal terminology, so write queries as you would
naturally ask a colleague.
- **Follow the thread.** PromptQL preserves your reasoning trail—feel free to pivot, rewind, or dig deeper without
losing previous steps.
- **Review the plan.** Each result is backed by a structured plan. Reviewing it helps validate how PromptQL interprets
your intent.
- **Use visualizations for communication.** When sharing results, use PromptQL’s built-in visualization capabilities to
highlight key insights clearly.

## Next Steps

It's great to be able to ask questions and get accurate, reliable responses. But, what if you could turn these into
automations? Check out how easy PromptQL makes it to [automate tasks](/automation.mdx) with the same level of accuracy
and reliability 🚀
2 changes: 1 addition & 1 deletion docs/how-to-build-with-promptql/_category_.json
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
{
"label": "Build with PromptQL",
"position": 1,
"position": 2,
"className": "basics-icon",
"customProps": {
"sidebar_pathname": "how-to-build-with-promptql"
Expand Down
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Loading