> ## Documentation Index
> Fetch the complete documentation index at: https://resources.athenaintel.com/llms.txt
> Use this file to discover all available pages before exploring further.

# When to use what on Athena

> Athena gives you chat, Skills, Agents, AOPs, Automations, User Interfaces, and Applications. Start with the simplest step that does the job and move up only when you need what the next step adds.

> Athena gives you building blocks from chat and saved Skills to Agents, AOPs (Agent Operating Procedures), Automations, User Interfaces, and full Applications. Start with the simplest step that does the job, and move to the next step only when you need what it adds. A method saved as a Skill, or an AOP that runs every week, often does the whole job without an app.

## Where do I start?

Each step on Athena adds a new capability and can use the steps above it. Start at the top of the table and move down a step only when you hit the reason in the last column.

| Step | Use it for | Move down when |
| - | - | - |
| **Chat** | A one-off question or task | You explain the same method a second time |
| **Skill** | Your method, written once and used whenever someone asks | Other people need to ask without setting anything up |
| **Agent** | A teammate your team can message in Teams, Slack, email, or text | The work needs to start without anyone asking |
| **AOP** | The same task done again and again: on a schedule, from a form, or when an email or file arrives. Athena records each run | You need fixed steps in a set order, approval between steps, or a spending limit |
| **Automation** | Fixed steps with approvals, retries, and a spending limit. It can call your AOPs and Agents | People need a screen to work in |
| **User Interface** | A dashboard or form over data you already have | People enter and change records all day, the app needs code running on a server, or people outside your workspace need to sign in |
| **Application** | A web app Athena builds on a Computer and publishes, with sign-in, saved versions, and a database | This is the last step. One Application is enough for an app your team or client relies on |

![A staircase of seven steps going down from left to right. Chat: ask a question or get a task done. Skill: save how your team does a task. Agent: give your team someone to message. AOP: run the same task again and again on its own. Automation: run set steps with sign-offs and a spending cap. User Interface: show your data on a simple screen. Application: build a full app that people sign in to. An arrow down the side reads go down a step only when you need more. Footer: each step can use the ones above it.](https://mintcdn.com/athenaintelligence-e46bc9d3/U4e92C1IcbO5Ytjz/images/build-guides/which-building-block-2.webp?fit=max\&auto=format\&n=U4e92C1IcbO5Ytjz\&q=85\&s=2245e9007fdd75c26bd31ee6862ae3b6)

You keep what you build as you move down the table: Agents and AOPs use your Skills, Automations call your AOPs and Agents, and Applications can start an AOP and read its result.

## How do I learn each step?

[Athena 101](https://resources.athenaintel.com/docs/athena-101.md) walks you through chat, Skills, Agents, AOPs, and Applications hands-on with sample files. You build a single project, a freight dispute desk, so taking the lessons in order shows how the steps fit together. Use the product guides in the table for Automations and User Interfaces.

| Step | Lesson |
| - | - |
| Chat | [Get a cited answer from your files](https://resources.athenaintel.com/docs/athena-101/03-cited-answer.md) |
| Skill | [Save your method as a skill](https://resources.athenaintel.com/docs/athena-101/11-save-a-skill.md) |
| Agent | [Give your team an agent, everywhere](https://resources.athenaintel.com/docs/athena-101/12-team-agent.md) |
| AOP | [Make it run every week](https://resources.athenaintel.com/docs/athena-101/13-weekly-aop.md), then [Build and trigger the exceptions AOP](https://resources.athenaintel.com/docs/athena-101/15-exceptions-aop.md) and [Approvals, failures and rollback](https://resources.athenaintel.com/docs/athena-101/16-approvals-failures.md) |
| Automation | [Automations](https://resources.athenaintel.com/docs/pillars/applications/automations.md) and [Automation approvals](https://resources.athenaintel.com/docs/pillars/applications/automation-approvals.md) |
| User Interface | [User Interfaces](https://resources.athenaintel.com/docs/pillars/applications/user-interfaces.md) |
| Application | [Build an AI app on your Computer](https://resources.athenaintel.com/docs/athena-101/18-build-ai-app.md), then [Publish it as an Application](https://resources.athenaintel.com/docs/athena-101/19-publish-application.md) |

Before people rely on what you built, read [Test it before you trust it](https://resources.athenaintel.com/docs/build-guides/test-before-you-trust.md).

## What does a full build look like?

Athena lets you combine these building blocks into a complete workflow, such as this invoice exception desk for a finance team.

| Part | What it uses | Why |
| - | - | - |
| How to judge a mismatched invoice | Skill | You write the team's method once and fix it in one place |
| Reading each mismatched invoice and suggesting why | AOP that uses the Skill | This step needs judgment, and Athena records each run |
| Nightly matching against the ledger | Automation that calls the AOP | Athena runs it on a schedule, in order, within a budget |
| Posting the approved fix | An approval step, then a connector to the ledger | A person approves before anything changes in the ledger |
| The screen where analysts work through exceptions | Application, built on a Computer | The team works in it all day, and changes go live only when you publish them |
| Invoices, exceptions, and their status | Athena-managed database | The app writes records and stops any exception from appearing twice |
| Three years of invoice history | Lakehouse | You can ask trend questions on a copy of the data |
| The monthly summary for the controller | Athena Sheets | People read it, adjust it, and check the sources |

You can run the desk before you build the screen: the Automation runs each night and the controller approves fixes in Athena. Add the Application when analysts need a dedicated place to work through the list.

## Which parts does an app use?

When you build an app on Athena, four questions help you pick the right parts for running code, doing the work, storing data, and reaching other systems.

| Question | Your options | How to decide |
| - | - | - |
| Where does it run? | User Interface, Computer, Application | Does it need a server? Do people depend on the link? |
| What does the work? | Chat, AOP, Automation, Agent | Should the AI choose the steps, or should you? |
| Where does the data live? | Database, spreadsheet | Does software write it, or do people review it? |
| How does it reach other systems? | Lakehouse, live connector, MCP server | Do you study a copy, or change the original? |

![Four questions, each with its parts. Where does it run? User Interface, a simple screen with no server; Computer, where Athena builds and tests; Application, the finished app people use. What does the work? Chat, one-off tasks; AOP, the same task again and again; Automation, set steps in a set order; Agent, a teammate people can message. Where is the data kept? Database, when the app saves records; Spreadsheet, when people check the numbers. How does it reach other systems? Lakehouse, a copy for analysis; Connector or MCP, read or change the original. Application, Automation, Database and Connector or MCP are highlighted as the parts a typical app uses.](https://mintcdn.com/athenaintelligence-e46bc9d3/U4e92C1IcbO5Ytjz/images/build-guides/which-building-block-1.webp?fit=max\&auto=format\&n=U4e92C1IcbO5Ytjz\&q=85\&s=b5d729c384f21ce2951d6e57a4cfd620)

The guides on [Application, Computer or User Interface](https://resources.athenaintel.com/docs/build-guides/application-computer-or-user-interface.md), [AOP, Automation or Agent](https://resources.athenaintel.com/docs/build-guides/aop-automation-or-agent.md), [Database or spreadsheet](https://resources.athenaintel.com/docs/build-guides/database-or-spreadsheet.md), and [Lakehouse or live connector](https://resources.athenaintel.com/docs/build-guides/lakehouse-or-live-connector.md) explain each choice.

## What should I use for my task?

| You want | Use |
| - | - |
| An answer to a one-off question | Chat |
| Your team's method, followed the same way every time | A Skill, in chat or in a specialist Agent |
| A teammate people can email, text, or message | Agent |
| Several specialists working on one job at once | Agent Team |
| The same judgment task, done again and again | AOP |
| Fixed steps with approvals and a spending limit | Automation |
| A dashboard or form over data you already have | User Interface |
| A place for Athena to write and test code | Computer |
| A link your team relies on, with saved versions | Application |
| Records an app creates and updates | Athena-managed database |
| A model or list that people review | Athena Sheets |
| The same analysis every week across many systems | Lakehouse |
| Today's value, or a change in the source system | Live connector or MCP server |

## What to plan for

Start with the lightest step that fits so you take on only the setup and maintenance you need. A Skill requires no releases, sign-in, or database management, whereas an Application includes all three.

Check with your admin before you plan around Applications, Automations, the lakehouse, or databases, as your Athena setup decides which of these are available.

Pick the step that covers your requirements, since each step has limits: use an Automation when you need approval steps between stages (an AOP has none), put a Skill inside an AOP or Automation when it needs to run on its own, and use an Athena-managed database when you need to stop duplicate records (a spreadsheet cannot). The linked pages explain each limit.

## Questions teams ask

> [accordion] Will Athena choose for me?
>
> Describe what you need and ask which steps fit. Athena explains the options and builds with the ones you pick. Naming the step in your request gives a more predictable result.

> [accordion] Can I move to another step later?
>
> Yes, and your data and instructions carry over. You can add a Skill to an Agent or an AOP, turn an AOP into an Automation draft, turn a User Interface into an Application, and load a spreadsheet into a database.

> [accordion] Do I have to go through every step?
>
> No. Start directly at the step that fits your job. If you already know people will enter records all day, start with an Application.

> [accordion] Is there a list of common setups?
>
> Yes. See [Reference patterns](https://resources.athenaintel.com/docs/build-guides/reference-patterns.md).

> [card] [See it work](https://athenaintel.com/demo/)
>
> Thirty minutes, your workflow, and you leave with the work product.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.