STORYPULSE AI

• 

• 

• 

• 

• 

Inspiration

A great movie idea is only the beginning. Before a film gets a greenlight, filmmakers and development teams need to understand audience interest, comparable films, market trends, creative risks, and how the concept fits into the current entertainment landscape.

I saw an opportunity to turn this traditionally slow, fragmented research process into an intelligent agentic workflow.

That led me to STORYPULSE AI — an AI development room designed to help filmmakers move from “Is this a good story?” to “Is this a story worth greenlighting, and why?”

Instead of simply generating another screenplay or movie recommendation, STORYPULSE focuses on decision intelligence for film development.

What it does

STORYPULSE AI takes a raw film concept and autonomously researches, analyzes, and evaluates it through multiple specialized agents.

A filmmaker provides a simple story idea along with details such as genre, language, and target audience.

STORYPULSE then:

• Breaks the concept into a structured research plan
• Searches the live web for relevant market and audience signals
• Finds comparable films and analyzes the competitive landscape
• Identifies creative, cultural, audience, and market risks
• Collects supporting evidence and sources
• Synthesizes the findings across multiple agents
• Generates a Greenlight Score
• Provides an actionable recommendation such as Greenlight, Greenlight with Changes, or Reconsider

The result is an evidence-backed Greenlight Dossier rather than a generic AI response.

How I built it

STORYPULSE is built as a multi-agent system using Gemini on Google Cloud and Parallel Search for live web intelligence.

The workflow follows:

Story Idea → Director Agent → Research Agents → Parallel Search → Evidence Validation → Risk Analysis → Greenlight Engine → Final Dossier

Agent architecture

Director Agent Understands the filmmaker's idea and creates the research plan.

Market Scout Agent Researches current market, audience, and industry signals using live web search.

Comparable Film Agent Discovers relevant films and identifies similarities, positioning opportunities, and potential competition.

Risk Analyst Agent Evaluates creative, cultural, audience, market, and production-related risks.

Greenlight Engine Synthesizes the research into weighted scores and produces the final recommendation.

Technology

• Gemini / Google Cloud — reasoning and agent intelligence
• Google ADK — multi-agent orchestration
• Parallel Search API — live web research and evidence retrieval
• Python + FastAPI — backend services and APIs
• Next.js + TypeScript — interactive frontend
• Tailwind CSS — UI system
• Recharts — analytical visualizations
• Firestore — project and research data
• Google Cloud Run — deployment

A key part of the implementation is that Parallel is used at runtime to retrieve fresh information for each analysis, rather than being mentioned only as a technology in the project documentation.

Challenges I ran into

One of my biggest challenges was making the system feel genuinely agentic rather than like a chatbot with multiple prompts.

I had to design clear responsibilities for each agent and determine what information should flow from one stage to the next.

Another challenge was dealing with the constantly changing nature of entertainment information. A static knowledge base would quickly become outdated, so I needed live research and source-backed evidence.

I also had to balance AI creativity with responsible decision-making. STORYPULSE should not confidently declare that a movie will succeed. Instead, it presents evidence, scores different dimensions, identifies risks, and explains why a recommendation was reached.

Finally, designing the experience was challenging because the underlying process is complex. I wanted filmmakers to see the intelligence happening without overwhelming them with technical details.

Accomplishments that I'm proud of

I'm proud that STORYPULSE transforms a simple movie idea into a structured AI-powered development workflow.

Instead of producing a single black-box answer, the system exposes the reasoning pipeline through an Agent Control Room, showing which agents are researching, analyzing, or waiting.

I'm especially proud of:

• Building a real multi-agent film development workflow
• Integrating live Parallel Search into the analysis pipeline
• Combining current web evidence with Gemini-based reasoning
• Creating an evidence explorer instead of hiding research behind a single answer
• Building a visual Greenlight Dashboard with explainable scores
• Detecting risks instead of focusing only on positive signals
• Turning research into an actionable Greenlight Dossier
• Designing STORYPULSE specifically around a real entertainment industry workflow

The goal was not to build another AI movie generator. I built an AI decision room for the people deciding which stories deserve to be made.

What I learned

I learned that building an effective agentic system is less about adding more agents and more about giving every agent a clear responsibility, useful tools, and structured outputs.

I also learned that live search becomes significantly more valuable when an agent knows exactly what it needs to find and why that evidence matters.

Another important lesson was that AI recommendations become much more useful when they are traceable. Instead of saying “this idea is promising,” STORYPULSE connects conclusions to evidence, comparable titles, risks, and measurable dimensions.

Most importantly, I learned that AI can be more powerful as a decision-support system than as a content-generation system.

What's next for STORYPULSE AI

The next version of STORYPULSE could become a complete AI development platform for studios, independent filmmakers, producers, and content teams.

Future capabilities include:

• Automated audience persona simulation
• Box-office and streaming performance intelligence
• Deeper regional and cultural analysis
• Budget-aware feasibility analysis
• Franchise and sequel potential analysis
• Script-level story diagnostics
• Automated pitch-deck generation
• Producer and investor-ready reports
• Historical trend analysis across genres and markets
• Continuous monitoring of market signals after a project is created

My long-term vision is simple:

STORYPULSE should become the intelligent development room that helps the entertainment industry discover not only which stories are exciting — but which stories are worth making.

Built With

• ai-agents
• cloud-run
• fastapi
• film-&-entertainment
• firestore
• generative-ai
• google-adk
• google-cloud
• google-gemini
• next.js
• parallel-search
• python
• tailwind-css
• typescript
• web-research