Reading the script before the studio does

A SaaS B2B business working with Cloud Combinator on AWS. The client is anonymised at their request.

The proof of concept had three focus areas, each essential to a script analysis the client could put in front of its customers.

A screenplay is ingested and broken into passages that fit the model's context, then a LangChain pipeline drives Amazon Bedrock, running Claude, through the scoring framework, prompting the model to assess each script against the client's defined qualitative and quantitative criteria. Real screenplays were used as test material to validate the approach across different genres and styles.

Cloud Combinator delivered a working proof of concept that ingests a full screenplay and returns a structured, criteria-based evaluation, demonstrating that a generative AI approach on AWS is a sound foundation for the client's script analysis and helping the team scope the platform and skills needed to take it forward.

• Whole-script understanding — The pipeline ingests a full screenplay and reasons across it, handling the length and conventions of real scripts.
• Framework-based scoring — Claude assesses each script against the client's qualitative and quantitative criteria, producing structured judgements rather than a loose summary.
• Consistent, comparable output — An explicit scoring framework means the same script yields a comparable evaluation each time, the basis of a data-driven service.
• Handover — The client received a validated MVP and a clear view of the platform and team needed to develop full script evaluation.

With AI-driven script analysis proven, the client is positioned to build automated screenplay evaluation into its platform, giving producers and creatives a faster, more consistent read on the scripts that cross their desks.