Orchard Core-integrated AI modules now available
You are surely well aware of all the AI tools out there. Perhaps, you also used them when working with Orchard Core, like generating code or asking questions of an Orchard Core-focused AI. Now, you have the option to manage your DotNest site with AI too! We at Lombiq, the company behind DotNest, use AI tools in our day-to-day software development workflows all the time. However, we also know that you have to be very critical of the output these tools generate, and you have to use them wisely, otherwise they'll actually slow you down. We also don't like if the programs we need for our work try to force us to engage with their AI features: Please, we'll use your AI stuff if it's indeed worthwhile, but let us decide for ourselves, thanks.So, we think about AI on DotNest in a similar way too: Here it is, we think it's cool, and feel free to use it if you'd like, but it's up to you. Keep in mind though, that this is currently experimental. Mike Alhayek, who is an exceptionally active Orchard Core contributor, is active in his CrestApps - Orchard Core project too. This contains a wealth of AI features as part of its Artificial Intelligence Suite. Mike has demoed these as progress went on, and you can find the videos under the Lombiq YouTube channel. And now you can use them right from your DotNest site! The features include support for all the major AI models that you can then use to manage the content or configuration of your Orchard Core site, or to simply chat with the model from the Orchard admin. Let's see how you can create a simple setup. Go to your site's admin area, to Configuration → Features, and enable the following features (you can just tick the checkbox to their left, and enable all at once with Bulk Actions in the top-right): AI Chat: To have a chat window on the admin. AI Connection Management: This contains the basics of setting up connections to hosted AI models like ChatGPT, Google Gemini, or DeepSeek models.Orchard Core AI Agent: Now, this is the actual secret sauce (well, as secret as something open-source can be). This is the feature that integrates the modules with Orchard Core, letting you manage your site with AI. And one of the features that support a concrete model family. Let's go with "OpenAI Chat" for this demo. You'll notice a new Artificial Intelligence item in the admin menu. Go to Provider Connections under it, then click Add Connection → OpenAI. Here, you need to set up a connection to an OpenAI model. We'll use DeepSeek for this example: Technical Name: DeepSeek. Display text: DeepSeek. Deployment name: "deepseek-chat". This is important, because this points to the specific model of the provider. This value points to DeepSeek-V3-0324 as of writing this. DeepSeek also accepts "deepseek-reasoner", which points to DeepSeek-R1-0528. You can check out the current possible values in the DeepSeek docs. Tick "Is default" for the sake of simplicity. Endpoint: "https://api.deepseek.com/v1". API key: You can create an API key on the DeepSeek Platform. Be sure to top up your account, even if it's just like $2. You can now add a Profile. This will be the configuration for the model, and thus affect how the model behaves and what it can do. Go to the Profiles menu, then click Add Profile → OpenAI. First, let's check out the Content tab (you should also see a Capabilities tab at the top): Display text: DeepSeek. You may notice a pattern here!Technical Name: DeepSeek. Deployment name: Default deployment. Profile type: Chat. This will allow us to chat with the model from the admin.Tick "Show on admin menu"Everything else is not mandatory, but you can customize how the model will work for you. Now you can go the Capabilities tab: Here is where the Orchard Core-specific, well, capabilities can be configured. These define what the models can do with the Orchard Core site, like managing content type and content items, turning features on or off, or managing users. Tick everything you're curious about and click Save!There will be a new admin menu under Artificial Intelligence now, called DeepSeek. This is where you can chat with the mode. Enjoy! For example, we can start by making the model praise DotNest! These models are great sycophants. But something slightly more useful is cross-checking what it can do. This should match the Capabilities we selected under the Profile: And then we can actually ask us to do things for us: There's a lot more things possible to do here, but you get the idea. It's the power of AI, but it can also work for you right there in your app.We also updated the SDK with the same modules, so you can use them to build your site locally too.What will you use AI for on your own DotNest sites?