Wrype's Case Study: AI Movie Script Insights & Analysis Platform

In our first of an on-going series of case studies, we take a look at how the Cloud Combinator team collaborated with the media and entertainment startup Wrype to deliver a ground-breaking AI/ML project in just two weeks. 

The Client: Wrype

Wrype is a significant emerging player in the field of media and entertainment. When the Wrype team got in touch, they explained how they wanted to explore ways of using artificial intelligence and machine learning to revolutionise movie script analysis and content creation in their industry.

By leveraging cutting-edge AI technology, the team at Wrype set out to create a user-friendly system to offer deep insights into movie scriptwriting. By doing so, they envisaged reducing the previously lengthy timescales and potentially inconsistent human processes involved in movie script development - thereby enhancing the quality and impact of media productions.

Why Cloud Combinator?

Wrype chose to collaborate with Cloud Combinator for this project owing to our expertise in AI/ML and our proven track record in deploying sophisticated machine-learning solutions in the media and entertainment industry. 

When we discussed our extensive experience handling similar projects and our deep understanding of AWS’s capabilities, the team at Wrype decided that these factors, along with our transparent way of working, made us an ideal partner for this ambitious project. Our mantra is all about getting working prototypes delivered at pace, so that the business can iterate both their technology and their ideas in near Realtime.

The Outline

In our initial discovery call, the team at Wrype explained it was essential that their movie script analysis tool could efficiently handle a broad range of movie scripts, identifying critical elements and then create meaningful development insights for the end-user.

While the generative feedback function of the platform was crucial, their end-product would also require an ‘admin interface’. It was important that this interface would make using the tool easy and intuitive for people who might not otherwise use AI/ML tools in their line of work.

Identifying Challenges

When an outline was agreed, the Cloud Combinator team began exploring the challenges that creating such a tool would present. 

Challenge 1: Creating the Data Pipeline

While there’s an understandable focus on the tool creating useful outputs, the inputting and handling of data into the tool also required careful consideration. The system would need to accept movie scripts in various formats before securely storing them and preprocessing them for effective AI analysis.

Challenge 2: Assessing LLM Capability

It was essential that the LLM chosen would be able to efficiently handle and analyse the submitted movie scripts with flawless accuracy. This would involve the capability to identify crucial movie script parameters which would then allow for the production of effective movie script evaluation.

Challenge 3: The Admin Interface

Whilst built on cutting-edge technology, simplicity and clarity are key when creating a user-friendly platform that bridges the gaps between the LLM, the data pipeline and the end-user. As such, a platform that facilitated ease of use, intuitive navigation and clear, understandable features was vital.

Developing the Tool

With the detailed outline completed and careful consideration of the challenges understood for creating the ML movie script Analytics Platform the Cloud Combinator team promptly moved on to development.

With significant experience and expertise around data pipeline development and best practices, we were able to quickly and effectively put together a solution that ensured the submitted movie scripts were handled according to a series of strict requirements. These requirements included security and storage - but also considering the preprocessing and transformation of data required by the LLM - and how the data would be integrated by the model.

In terms of the foundational model itself, Anthropic’s Claude V2 stood out as being the perfect fit. The latest iteration of the Claude LLM has extremely advanced understanding capabilities, ensuring nuance and accuracy. What’s more, Claude V2 is unparalleled in terms of the amount of data it can handle without losing context - essential for potentially long and complex script inputs.

We also integrated the development platform LangChain alongside Claude to enhance the tool's capabilities. LangChain's versatility in integrating distinct models, and its ability to create modular, customizable solutions, proved invaluable. Its seamless integration with Claude V2 enabled us to tailor our solution specifically to the project's requirements, ensuring flexibility and the effective handling of complex script data with advanced AI-driven analysis.

At the heart of the movie script analysis project was Amazon Bedrock. Not only did Bedrock allow us to compare the suitability of foundational models and enhance the solution with LangChain - it was also the perfect platform in which we could create an effective, user-friendly admin interface for the tool.

Collaboration and Visibility

When teams collaborate, transparency is a non-negotiable. This was an especially important factor when working with Wrype when considering that the timescales involved were to be measured in days, rather than weeks or months. 

By laying out every step and every development of the project in real time in a GitHub repository, the Wrype team had immediate oversight throughout and open lines of communication with the team here a Cloud Combinator. 

Added Value

When we connect with a client, it’s normal for there to be a strong focus on the here-and-now requirements for a proof of concept or a minimum viable product. While this is essential for a quick POC/MVP delivery, we always work behind the scenes to ensure that system architecture will allow for the practical ongoing development of a project. This was an important part of our collaboration with Wrype.

By using Bedrock to create an effective interface and integrate the right LLM for the movie script analysis tool, we could be 100% confident that every part of the platform could be easily enhanced to potentially handle even more advanced movie script analysis features in the future.

Support and Funding

While confidence in Cloud Combinator’s ability to do the technical work required for the movie script analysis tool was obviously important for the team at Wrype, our position as an AWS Advanced Partner meant the team could also be 100% confident we were leveraging all possible funding avenues and exploring cost-saving opportunities through every step of the project too.

Client Satisfaction

Wrype’s feedback on the MPV delivered and the partnership with the team here at Cloud Combinator has been extremely positive - paying special praise to our swift 2-week delivery, enthusiastic engagement, and our focus on ensuring future scalability for the project.

“Working with Cloud Combinator has been a godsend for us. They have been professional, efficient, honest and creative. Wrype would not be where it is today without their team, and we look forward to continuing to work closely with them in the future.” 

Igor Mongulu Co-founder