-
-
-
CineVerdict turns film concepts into evidence-driven greenlight decisions using Gemini and Parallel Search.
Inspiration## Inspiration
Film and entertainment decisions often combine creative instinct with fragmented market research. CineVerdict was built to turn that process into an evidence-driven, agentic workflow. Instead of simply generating an opinion about an idea, CineVerdict researches it, evaluates the evidence, identifies uncertainty, and produces an actionable greenlight recommendation.
What it does
CineVerdict is an agentic film intelligence system that evaluates entertainment concepts through a coordinated multi-agent workflow.
A user submits a film or documentary concept. CineVerdict then orchestrates specialized agents to research the concept, analyze market signals, evaluate production and risk considerations, and synthesize the findings into a structured verdict.
The final result includes:
- A GREENLIGHT, MODIFY, or PASS recommendation
- Confidence level
- Evidence ledger
- Decisive reasons
- Unresolved uncertainties
- Recommended next actions
How we built it
CineVerdict is powered by Gemini on Google Cloud and integrates Parallel Search for evidence retrieval.
The workflow uses specialized agents for direction, research, market analysis, production/risk evaluation, and final decision synthesis. Parallel Search provides external research evidence that is carried through the agent workflow rather than simply being mentioned in the interface.
This architecture allows CineVerdict to separate evidence gathering from reasoning and final decision-making, producing a traceable recommendation instead of an unsupported AI opinion.
Challenges we ran into
One of the biggest challenges was maintaining evidence provenance across a multi-agent workflow. Research results needed to remain connected to the conclusions that depended on them.
We also focused on deterministic orchestration, graceful handling of uncertainty, and ensuring the final verdict clearly distinguishes evidence, reasoning, unresolved questions, and recommended actions.
Accomplishments that we're proud of
We built a complete end-to-end agentic workflow rather than a single prompt or chatbot.
CineVerdict can take an entertainment concept through research, market analysis, risk evaluation, and final decision synthesis while maintaining an evidence ledger throughout the process.
We are especially proud of integrating Parallel Search directly into the runtime research workflow and using Gemini to coordinate specialized reasoning across the system.
What we learned
Building useful agentic systems requires more than connecting several models together. Agent responsibilities, evidence provenance, failure handling, and decision contracts need to be explicit.
CineVerdict demonstrated how agentic AI can augment creative decision-making without replacing creative judgment: research and structured reasoning help decision-makers understand why an idea should move forward, change, or stop.
What's next for CineVerdict
The next step is expanding CineVerdict into a broader pre-production intelligence platform with richer market intelligence, comparable-project analysis, audience signals, configurable decision criteria, persistent project histories, and collaborative workflows for creative teams.
The long-term vision is simple: help great stories reach better-informed greenlight decisions.
Log in or sign up for Devpost to join the conversation.