AI Provider for amazee.ai

This plugin integrates amazee.ai private AI hosting with WordPress AI features, enabling secure, sovereign, and GDPR-compliant AI capabilities on your site. This plugin requires WordPress 7.0 or newer.

From within WordPress

1. Visit Plugins > Add New.
2. Search for AI Provider for amazee.ai, then click Install Now and Activate.
3. Follow the Configuration steps below.

Manual installation

1. Download the plugin zip, then upload it via Plugins > Add New > Upload Plugin. Or extract it into wp-content/plugins/ yourself.
2. Activate the plugin through the Plugins menu in WordPress.
3. Follow the Configuration steps below.

Composer For sites managed with Composer: composer require wpackagist-plugin/ai-provider-for-amazee-ai Released versions bundle their own autoloader, so there is no build step. Only if you install from a Git clone do you need to run composer install --no-dev inside the plugin directory. Configuration

1. Obtain your credentials:
2. Log into your account at my.amazee.io to obtain your endpoint URL and API key.
3. Set the endpoint URL:
4. Navigate to Settings > amazee.ai (/wp-admin/options-general.php?page=ai-provider-for-amazee-ai) and enter your endpoint URL, for example https://llm.us103.amazee.ai where us103 is a US region. Regions are also available in the UK, Germany, Switzerland, Australia and more.
5. There is no default endpoint: your API key only works with the region it was issued for, so copy the exact URL from my.amazee.io.
6. Store the API key:
7. Navigate to Settings > Connectors (/wp-admin/options-connectors.php), locate the amazee.ai connector and enter your API key.
8. The Settings > amazee.ai screen then shows the connection status and the models available to your account.
9. Alternatively define AMAZEE_ENDPOINT_URL and AMAZEE_LLM_TOKEN constants in wp-config.php (or set the AMAZEEIO_API_KEY environment variable) and skip the UI entirely. The pre-1.4 url|token credential format keeps working.
10. Enable AI experiments (optional):
11. To actually use the connector, install and activate the official AI Experiments plugin.
12. Navigate to Settings > AI Experiments (/options-general.php?page=ai-experiments)
13. Select »Enable Experiments« and Save.

1.5.1

• Requests to your amazee.ai region endpoint now use a 30 second timeout instead of the 5 second WordPress default, so model discovery and direct AI client calls (e.g. image generation) no longer time out early. Requests made through WordPress core's AI features keep core's own timeout.

1.5

• Embedding generation: models the endpoint reports in embedding mode are exposed with the embedding generation capability (php-ai-client 1.4+ or the WordPress AI plugin 1.5 SDK overlay).
• Text to speech: models in audio_speech mode are exposed with the text-to-speech capability, so the AI plugin's Text to Speech experiment turns itself on in regions that offer such a model.
• Audio input: chat models reporting supports_audio_input accept audio files in prompts.
• Security: the endpoint URL must use https, so the API key is never sent unencrypted. The settings screen refuses plain http URLs, and requests refuse them whichever way the URL was configured (constant, setting or credential).
• Accessibility: connection status uses core notice styles (the green status text failed WCAG AA contrast), the endpoint URL field is linked to its description, and the my.amazee.io link announces that it opens in a new tab.
• Settings > amazee.ai shows each model's capabilities (text, vision, audio input, image generation, embeddings, text to speech) as discovered from the endpoint.

1.4

• New Settings > amazee.ai screen for the endpoint URL, with a connection check listing the models available to your account.
• The amazee.ai connector on Settings > Connectors now takes the API key on its own — no more url|token pipe format needed. Existing url|token credentials keep working.

1.3

• Expose image generation: models the endpoint reports with the image_generation mode are now advertised with the image generation capability, so features such as the AI plugin's image generation turn themselves on in regions that offer such a model.
• Prefer the region's chat and text_to_image aliases when a feature has no explicit model preference, instead of whichever model the catalog happens to list first.

1.2.1

• Rewrite the installation instructions for the WordPress.org release: install from Plugins > Add New, or upload the zip. Composer is now a note for Composer-managed sites rather than the only documented path.

1.2

• Integrate with the WordPress 7.0 Connectors screen: the provider now declares API-key authentication, so the credential is stored and resolved by the core AI Client and handed to the provider through it. The plugin never reads the stored credential itself.
• The credential may include the endpoint: https://llm.<region>.amazee.ai/v1|<token>.
• Send an X-Amazee-Client header identifying the plugin and version with every API request.

1.1

• Support Composer-based installs that provide a site-wide autoloader.
• Filter request parameters against each model's supported OpenAI parameters.
• Cache the model catalog for 12 hours per endpoint.
• Show an actionable message when the amazee.ai budget is exceeded.

1.0

• Initial release.