Configuring AI Providers
To protect your data, AI-assisted development is disabled by default and needs to be configured independently in each of your Wing Pro projects before it will work. You also need to explicitly approve uploading content to the AI provider. Unless both these steps are taken, the AI integration will not work and no content will ever be uploaded to the AI provider. This design allows you to use AI in some projects, while avoiding it entirely in projects where privacy or other concerns prevent its use. See also AI Assisted Development Privacy and Data Protection.
Wing currently supports OpenAI, Gemini, Grok, Perplexity, Claude, Mistral, and Deepseek as AI providers, as well as any other provider that implements an OpenAI compatible completions API. In order to use AI in Wing, you must have your own account and API key with your selected AI provider. Some AI providers offer a free usage tier, but in many cases you will need to purchase API credits in order to gain access to more capable AI models or to lift rate limits. If you find that you are getting poor results, this may be due to the limitations on the freely available models. AI is very compute and energy intensive so it's not surprising that the free models are rather limited in their abilities.
Wing will prompt you to configure an AI provider the first time you try to use any AI feature, or you can do it from Manage Providers in the AI menu.
Provider Configuration
To create an AI provider configuration, select Manage AI Providers from the AI menu and press the + icon.
On the Provider tab of your AI provider configuration, you will need to enter a name for the configuration, select the AI provider to use, specify the API key, and select the language model to use. Depending on the model chosen, you should also configure the currency and token costs to match what you are being charged by your AI provider, so that you can keep track of the cost of requests that you make from Wing.
You will need to log into your AI provider account to generate your API key. The key can be placed into a file (the default location in your User Settings directory is given in the dialog) or enter it directly under Use Key. We recommend using a file so the the key is less likely to be shared accidentally and so you can more easily reuse it in other AI provider configurations. If you do enter your API key directly, it will be written to the .wpu user and machine specific branch of your project, which typically should not be shared because it is intended only for the local environment.
Once you save your first AI provider configuration, Wing will start to use it. However, it will not upload any data to the provider until you have also approved content uploads. You will be asked to approve uploads the first time you use the AI provider, or you can select Approve Uploads to OpenAI in the AI menu. Once this is done, you can use Revoke Upload Permissions in the AI menu at any time to prevent further content uploads.
Context Configuration
The Context tab of the AI provider configuration allows you to configure thresholds for how much context Wing will upload with your AI request. The values used here depende on the capabilities of the AI language model that you have selected:
- Max Context Unit Size (KB) sets the maximum number of kilobytes of code from any given file that may be submitted with an AI request.
- Max Total Context Size (KB) sets the maximum size in kilobytes of all context that may be submitted with an AI request. This includes context from the current file, any manually specified context, and files selected by automatic context options.
- Max Upload Size (KB) sets the maximum total size in kilobytes of all files that may be uploaded to the AI provider at any given time.
- Max Upload File Count sets the maximum number of files that may be uploaded to the AI provider at any given time.
The maximum sizes to use here are governed by the capabilities of your AI provider and language model, the specific rate limits imposed by your AI provider account, and the cost of requests. Note, however, that the upper limit may never be reached if your files are small and/or you have not enabled the use of additional context.
It may take some experimentation to find the best value to use, where enough context is provided for quality results without overwhelming the AI provider or running into your account's rate limits.
Testing the Configuration
To test your configuration, open the AI Chat tool from the AI menu, enter "Can you write a Python hello world?", and press the Send button. You should see a Waiting message appear at the bottom of the AI Chat tool. When the operation completes the result will appear in the AI Chat tool.
If this doesn't work, select Show Console in the AI Chat tool's Options`` menu. The console shows everything Wing sends to the AI provider and all responses and error messages.
Section Contents
