Tutorials

Tutorial: Set Up AI Settings and Personal API Connections in PromptPlan

Learn what AI settings do in PromptPlan, how to connect OpenRouter, OpenAI, or Claude, how default models work, and how your API keys are stored.

Updated 2026-03-22

#tutorial#ai settings#api keys#providers

Goal

Understand what the AI settings page controls and set up your own model connections correctly.

This tutorial explains:

  • what AI settings are for
  • which providers PromptPlan supports
  • how to connect your own API key
  • how saved connections affect model pickers
  • how the default model works
  • how PromptPlan handles your key securely

Step 1: Open AI settings

Go to /settings/ai.

This page controls your personal AI connections for PromptPlan.

It does not replace PromptPlan’s free models. Those stay available unless a specific feature requires a compatible model that only appears through your own connection.

Step 2: Understand the three main areas

The AI settings page is split into a few practical sections:

Default model

This lets you choose the model PromptPlan should prefer by default when a model picker needs an initial selection.

This is useful if:

  • you usually want one provider
  • you prefer a specific model family
  • you want runs and editor actions to start from your own preferred default

Saved connections

This shows every API connection you have already added.

Each saved connection includes:

  • the provider name
  • your connection label
  • a masked key preview
  • the last validation time when available
  • the most recent validation error when something failed

Add connection

This is where you connect a new provider key.

You choose:

  • Provider
  • API key
  • Connection label (optional)

Then click Connect API key.

Step 3: Choose a provider

PromptPlan currently supports:

  • OpenRouter
  • OpenAI
  • Claude

Use the provider dropdown to select the service your API key belongs to.

Important:

  • if you use an OpenRouter key, choose OpenRouter
  • if you use an OpenAI key, choose OpenAI
  • if you use an Anthropic/Claude key, choose Claude

Choosing the wrong provider can cause validation to fail even if the key itself is valid.

Step 4: Paste your API key

Paste your provider key into the API key field.

PromptPlan will validate the connection when you save it.

If validation fails, check:

  • whether the provider matches the key
  • whether the key was copied completely
  • whether the key is active in your provider account
  • whether the provider account has the necessary access enabled

Step 5: Add an optional connection label

The Connection label field is optional.

Use a label when you want to distinguish connections such as:

  • OpenRouter - personal
  • OpenAI - team
  • Claude - research

If you leave it blank, PromptPlan names the connection automatically based on the provider.

Step 6: Connect the key

Click Connect API key.

When the connection succeeds:

  • it appears in Saved connections
  • PromptPlan validates it
  • compatible models from that connection become available in model pickers

This affects model selection across the product, including places where PromptPlan lets you choose a model for templates, workflows, editor actions, or chats.

Step 7: Set your default model

After you connect a provider, open the Default model section.

Choose the model you want PromptPlan to default to when a picker needs a starting value.

This does not force every run to use one model forever. It sets the default selection, while specific prompts, templates, workflows, or other surfaces can still use different models when needed.

Step 8: Review saved connections

Use Saved connections to audit what is active.

This helps you:

  • confirm which providers are connected
  • see whether a connection was validated
  • spot errors quickly
  • remove outdated keys

If a connection is no longer needed, remove it from this section.

How PromptPlan uses your key

Your connected key is used only for your own AI requests in PromptPlan.

That means PromptPlan uses it to send model requests initiated by your own activity in the product, such as:

  • working in supported editor AI actions
  • running supported AI-powered flows
  • using model pickers that rely on your saved connection

PromptPlan does not expose your key back in the UI beyond a masked preview.

How PromptPlan secures your key

PromptPlan stores user API keys encrypted.

In practical terms:

  • your full key is not shown back to you after saving
  • only a masked preview is displayed in the UI
  • PromptPlan uses the saved encrypted key only to perform your own requests
  • the key is tied to your own account settings, not shared globally

Common setup mistakes

Provider mismatch

The most common issue is selecting the wrong provider for the key.

Example:

  • an OpenAI key entered under OpenRouter
  • a Claude key entered under OpenAI

Expecting a connection to replace all models

Connecting your own provider adds models and can influence defaults, but it does not mean every surface instantly becomes locked to that provider.

Forgetting to choose a default model

If you connect a provider but never set the default model, PromptPlan may still begin from another available default selection depending on the feature.

For most users, a clean setup looks like this:

  1. connect one provider first
  2. confirm it validates successfully
  3. choose a default model
  4. add a second provider only if you have a clear reason
  5. label connections clearly if you keep more than one

Final takeaway

AI settings exist to give you controlled access to your own providers inside PromptPlan without losing visibility over what is connected and what model is selected by default.

If you treat the page as:

  • your provider connection manager
  • your default model selector
  • your connection audit log

then the whole screen becomes much easier to use.

Related reading

PreviousTutorial: Customize a Starter Workflow in PromptPlan
NextKeyword Research Mastery