Leveraging GenAI for Enhanced Content Creation in AEM
There is no doubt that our generation has to go hand in hand with generative AI capabilities. It brings in efficiency in day to day work by super accelerating the launch of new products and features in any kind of an organization.
Today here in this blog, we delve into the ways we can use external AI models within AEM to bring efficiency into the day to day work of an AEM content-author or AEM developer.
DALL-E 2/3 from openAI
DALL-E is a transformer language model just like GPT3. We all are now familiar with chatGPT and its capabilities. Now how do we bring these capabilities into AEM ?
Adobe Sensei
Obviously, Adobe has its own sensei which brings in AI capabilities within Adobe’s creative cloud applications, Adobe Express and Adobe firefly. All these are licensed products and the extreme AI capabilities of firefly can’t be utilized within AEM Assets out of the box as of yet and we don’t have a timeline when it will be available. By extreme capabilities I mean to say the text to image generation, text to image modification features. So what does a customer do with only AEM Sites or AEM Assets license ?
Here is our solution to it.
AEM custom component
Here is a custom component in AEM to give the customer a GPT like interface and environment to get indulged in an AI enabled content creation and modification process right within AEM.
Image Generation
Our custom component interacts with the DALL-E model to generate images from text prompts and then allows the user to save the image within AEM DAM.
Image Modification
The second stage of it is to have the capability to fetch an image from AEM DAM directly and customize it with text prompts. The modified image will get saved in DAM.
Flexibility of choosing backend Models
In the above ways as we mentioned, there is also going to be flexibility with the end user to choose what models would the user like to use. With the availability of public LLMs ( Large Language Models ) we can leverage more and more genAI capabilities into whatever way we want them to be.
Background to the above processes
Some of the organisations hosting the TLMs (transformer language models) do have a user subscription model to interact with their models for genAI capabilities through authenticated API requests. This is how they become immensely popular day by day and we can also leverage this opportunity to ease our daily work in AEM.
References
