Foundation / First setup

The Beginner AI Agent Setup Checklist

A productive first setup is intentionally small. The goal is not maximum autonomy. The goal is a workspace you understand, a budget you control, and three workflows that produce inspectable drafts.

01

Define the job before choosing tools

Write the primary outcome in one sentence and name the recurring tasks that support it. A business owner may need research and draft preparation; a freelancer may need meeting follow-up and client delivery checks.

Choose one workspace for the experiment and keep sensitive customer, health, financial, legal, and credential data out until the environment and policy have been reviewed.

  • One primary outcome
  • One bounded workspace
  • Three candidate workflows
  • No sensitive data by default
02

Start with the least permission

Use advisory and read-only work first. Add local drafting only when you can inspect the files. Keep external sending, publishing, spending, deleting, and deployment behind explicit human approval.

Do not paste API keys into chat. Store credentials through the host-supported environment or credential system only after the integration is actually required.

  • Read before write
  • Draft before send
  • Preview before publish
  • Approve before spend or delete
03

Set a zero-dollar default

Use models and subscriptions you already have before adding paid APIs. Record which tasks can use a low-cost model and which require stronger reasoning or code review.

Set a monthly external API ceiling of zero until you choose otherwise. When paid access becomes useful, add a defined limit, usage review, and alert rather than an open-ended key.

  • Existing access first
  • Zero-dollar API default
  • Task-based model routing
  • Monthly usage review
04

Specify the first three workflows

For each workflow, write the trigger, inputs, steps, tools, output file, approval gates, and acceptance checks. Run one low-risk example and review the result instead of automating the recurrence immediately.

Agent Ready packages these decisions into four working documents. The documents remain useful even if you later change platforms because they describe your operating rules rather than one vendor interface.

  • Research or decision brief
  • Planning or status draft
  • Role-specific workflow
  • Human review after every first run
Put it to work

Build the controlled foundation first.

Agent Ready turns these principles into four working documents before it asks you to automate anything.

Get Agent Ready Free
Continue the field manual

Related guides