Although we usually say that we “talk” to ChatGPT, the truth is that what we do is write to the chatbot and read its replies. This nuance may sound like a small detail, but the truth is that it is not. Services like WhatsApp and the like, and earlier SMS, as well as email and so on, have made text communication much more common than it was 20 or 30 years ago, but still, for many people, the most comfortable system for communication is verbal. , especially for conversations similar to those suggested by many chatbots.

On the other hand, and as I predicted when OpenAI opened up the ChatGPT API, We’ve seen all kinds of designs come up which are trying to add some added value to what chatbot technology already offers, in a progression that we can be sure will continue at least in the short term. Forefront personalities and Poe’s own chatbots are just two examples of an ecosystem that’s expanding by the second, making the offerings not only more abundant, but also much more diverse and thus more interesting.

What is good for users compels the creators of these implementations try harder to find original designswith which to differentiate yourself from this growing competition, whether it offers access from different channels, integration of several chatbots into one service, adding their functions to other applications and services, etc., which in turn serves to significantly expand the scope of ChatGPT, derivatives and the like.

One of the latest and most interesting examples I found in this regard is an iOS app that is also available through the web, albeit to a more limited extent. It’s called Call Annie and it offers us a video call with a chatbot who as you can tell is called Annie. As you can see in the images above the previous paragraph, but only if you have an iPhone 12 or higher, you will be able to talk to this AI-based assistant and have a conversation that provides a reasonably natural experience.

Call Annie web interface.

for this purpose Call Annie combines several technologies. The basic one is of course the generative model of artificial intelligence that generates answers from our calls, but in addition we find another artificial intelligence that is responsible for converting our voice to text. And on the other hand, also those responsible for the animation of the model, originally created with Midjourney, so that its response, transformed into a text-to-speech solution, synchronizes it with the facial expressions of the animation. The set therefore has the most interesting technical invoice.

As I said earlier, the video conferencing feature through the app is only available for iPhone 12 and above because the image animation is created by the Neural Engine. However, for those who have previous models, the same function is offered that we can find through the web, in which the conversation takes place exclusively by voice.

Standard in this respect Annie speaks English, although you can speak Spanish with her, which also makes it a very interesting possibility to practice the language. It’s true, of course, that you can ask him to answer you in Spanish, but since his text-to-speech solution is exclusively for English, the result will be a bit… disturbing.