Alexa, Find My Movie: A ChatGPT-Powered Revolution in OTT Apps

Who can resist the charm of a voice-activated search engine for an OTT app? There's something undeniably cool about it, isn't there? ?? A few weeks back, I wrote an article about my little experiment with ChatGPT, where I used it to build a search engine for an OTT app. The response was quite positive, with several folks reaching out to me, curious about the details and expressing how fantastic it would be to incorporate voice functionality into the mix.

So, I thought to myself, why not? Let's add a bit of voice magic to the mix. But what could I use to make this happen? Phone with Siri, a voice command button on a text box?

One evening, I was sitting on my sofa, half-watching a rather boring movie. My mind started to drift, and my eyes landed on my Alexa device, just sitting there on the side table. It was a simple moment, but it sparked an idea. With an Alexa in every room of my house, it seemed like the obvious choice. It was a lightbulb moment - if I had an Alexa-powered search engine, I could just ask Alexa to find me a better movie. It felt like a natural solution, almost like it was meant to be.

Alexa comes with this nifty feature called skills, which are essentially small modules or plugins that you can use to extend Alexa's functionality. Although I've used Alexa skills before, I've never actually developed one. So, I decided to dive into the Alexa documentation for developers. It turns out, Alexa skills can be developed using JavaScript, and it's actually quite straightforward. So, I decided to give it a shot.

With Alexa as the input, I needed to figure out the rest of the setup. I decided to build a simple web API in .Net for the backend. The API's job would be to receive the input from the Alexa skill, connect to the ChatGPT API, and filter the movies, similar to what I explained in my previous article. The final step was to figure out how to send the payload server-side to the client app to visualise the search result.

For this, I decided to use a technology called websockets, which allows an app to have a bi-directional persistent connection with a server. In this case, I can send anything I want from the server to the client, without the client having to ask for it. Specifically, I decided to use a Microsoft Azure service called SignalR.

With all the pieces in place, I started the implementation, beginning with the Alexa Skill. I developed the Web API and implemented the connection with SignalR. I also had to extend the client to add the code to connect to SignalR and render the result.

To test the result, I hosted the Alexa Skill in the Amazon Alexa developer portal, SignalR is an Azure managed service, and I decided to host the Web API and the React client app on my local Mac. To expose the web API to the internet, I used a handy little utility called NGROK, which essentially creates a tunnel with port remapping capabilities.

As usual, for the tech-savvy, here's a sequence diagram that explains the sequence of actions:

And here a working video that shows the process

The solution worked quite well. Alexa did a fantastic job of translating the command, understanding it, and sending it to my skill. ChatGPT also did a great job of understanding what to filter, providing only the search criteria and nothing more. ChatGPT is also great at understanding what you want, even if there are typos in the request.

Now, let's take a moment to consider the real-world implications of this proof of concept. Imagine for example, being able to simply tell your OTT app what you're in the mood for, and having it instantly provide you with a selection of movies or shows that match your request. It could revolutionise the way we interact with our OTT apps, making them more user-friendly and personalised.

While voice-activated search isn't a new concept, it's traditionally been used to simply convert spoken words into text for a standard text-based search. However, the combination of Alexa and ChatGPT in my experiment offers a whole new level of advanced search capabilities. It's not just about finding a movie with a specific title or actor anymore. With this setup, you can perform nuanced searches based on mood, scenarios, or even complex combinations of criteria. For instance, you could ask for all adventure movies directed by a specific director, featuring a certain actor, and with a touch of comedy. It's a more natural and intuitive way of searching, much like asking a knowledgeable friend for movie recommendations.

So, what's the point of all this? From my perspective, it's just for fun and to prove to myself that I can achieve the desired result. From your perspective, I guess it's a glimpse of what's possible to make something like an OTT app more enjoyable to use. I'm a strong believer that in the next 4-5 years, AI will be everywhere, OTT apps included, and I'm trying to imagine the benefits that AI can bring to an OTT app to improve it and make it more enjoyable.

In conclusion, this experiment serves as a testament to the potential of AI in enhancing our everyday experiences, even in areas we might not initially consider, like our favourite OTT apps. It's a fascinating glimpse into a future where AI and voice technology make our interactions with technology more natural and intuitive.

I have lots of new ideas that bring together OTT and AI that I want to test (time permitting). But if there's something specific you'd love to see implemented, do let me know in the comments. If it's within my skillset/knowledge, I'll try to experiment with it.

If you enjoyed this article, consider following me here on LinkedIn. I talk about leadership, management, and technology. So, what do you reckon? Would you fancy having something like this in your favourite OTT app? Remember, the future of OTT apps is not just about what we watch, but also how we choose what to watch.

#voicesearch #alexa #chatgpt #ott #ottplatforms #searchengines #ai #innovation #voicetech #naturallanguageprocessing #streamingmedia #streaming