Blog
Build production-ready AI agents in Agentic Cinema: The Blockbuster Hackathon
Participant resources

Build production-ready AI agents in Agentic Cinema: The Blockbuster Hackathon

Build an agent with Google Cloud and one of 5 star-studded partners: IBM, Grafana Labs, Parallel, ClickHouse, or Replit.

Text reads "Google Cloud's Agentic Cinema Blockbuster Hackathon" next to the Google Cloud logo

Step onto the studio lot! Google Cloud is teaming up with an all-star partner ecosystem for Agentic Cinema: The Blockbuster Hackathon, a challenge for builders ready to turn enterprise complexity into a summer blockbuster.

This hackathon brings together Google Cloud and five technology partners: IBM, Grafana Labs, Parallel, ClickHouse, and Replit. Together, they give your agents the real-world capabilities they need to take on production-scale problems.

Key details

  • Submission deadline: September 9, 2026 at 2:00 p.m. PDT
  • Total prize amount: $75,000
  • What to build: A functional agent, powered by Google Cloud and Gemini Enterprise Agent Platform, that integrates one partner's product or MCP server to power a real media and entertainment workflow.
  • What to submit: A URL to your hosted project, a ~3-minute demo video, your public code repository (GitHub, GitLab, or Bitbucket) with a complete open-source license, your selected partner track, and your completed Devpost submission form. Your repo needs to show actual runtime use of both Google Cloud and your chosen partner's service. (Read the Official Rules for more details.)

Good to know: Projects can only use Google Cloud AI tools and your chosen partner's built-in AI features—no other AI models, agent frameworks, or AI APIs, including the tools you use to write your code during development. Gemini CLI and Gemini Code Assist are your safe path for AI-assisted coding throughout this hackathon.

About the prize structure: This hackathon features five identical prize pools for each partner track. Each track awards a first-place ($7,500), second-place ($4,500), and third-place ($3,000) winner.

Who should join

Whether you see yourself as the Director architecting an autonomous agent network, the Technical Producer wiring up secure data pipelines, or the Studio Head enforcing governance across a multi-agent workflow, there's a place for you here. 

What you'll build

Projects need to solve a real bottleneck across the entertainment and media value chain, whether that's the workflow of a filmmaker, screenwriter, studio crew, or fan. 

Your project doesn’t have to be a tool for a two-hundred-person production team. You can also build for the solo YouTuber or TikTok creator who writes, shoots, edits, and posts alone.

Lean on Gemini's multimodal and generative capabilities as you build: generate VFX concept art and storyboards with Imagen 3, score a scene with Lyria 3, voice dialogue with Gemini TTS, or transcribe and caption raw footage. 

Whether you're orchestrating an autonomous agent network, wiring up a secure data pipeline, or enforcing governance across a multi-agent system, your agent should do more than answer questions. It should plan, act, and get the job done.

How to build

  1. Register for the hackathon on Devpost: Click ‘Join hackathon’ to get started.
  2. Set up your environment: Start with the Gemini Enterprise Agent Platform API Setup, the control panel for your Google Cloud agent configuration. Sign up for a no-cost trial at cloud.google.com/free, or request $100 in hackathon credits through this form by August 31st.
  3. Pick your path: Use the Agent Builder Guide for a low-code build, the Gemini Enterprise Agent Platform SDK for Python for custom agent logic, or the Agent Starter Pack to initialize a custom backend. 
  4. Choose your partner: Pick the partner platform that holds the data or workflows your agent needs to interact with—IBM, Grafana Labs, Parallel, ClickHouse, or Replit.
  5. Solve a real workflow: Show off a deterministic, multi-step agent that integrates your chosen partner’s product or MCP and solves genuine enterprise friction in a media and entertainment context.
  6. Submit your project: Share your hosted project URL, your open-source code repository, and a ~3-minute demo video showing your agent in action.

Now let's take a closer look at each partner and what they bring to the hackathon.

{{callout-1}}

IBM

IBM is a leading provider of global hybrid cloud, AI, and consulting expertise, helping clients in more than 175 countries streamline operations and capitalize on their data. For this track, IBM is bringing IBM Bob, an AI-assisted development environment, into your build process. See all the details on the IBM partner track.

Some ideas to spark your thinking:

  • A production scheduling agent that helps coordinate crews, equipment, and locations, and adapts automatically when plans change
  • A content pipeline agent that uses IBM Bob to accelerate development and Confluent to stream real-time updates as footage moves through post-production
  • A rights and licensing agent that cross-references usage data to flag compliance issues before they become costly

Requirements

Your project must be built using IBM Bob as part of the development process. Use of Confluent is optional but strongly encouraged to power real-time and event-driven workflows. Strong submissions should also demonstrate how AI meaningfully improves a workflow, enhances decision-making, elevates a customer experience, or drives a measurable operational outcome.

IBM resources

Grafana Labs

The Grafana stack brings metrics, logs, traces, profiles, dashboards, and alerting together in one place, and the free Grafana Cloud tier covers everything you need for this hackathon.

Give your agent live production context with the Grafana Cloud MCP server. Pair it with Grafana Cloud AI Observability to track your own agent's LLM calls, token cost, latency, and MCP tool activity while you build. See all the details for the Grafana Labs partner track.

Some ideas to get you started:

  • An agent that investigates a firing alert, correlates Loki logs with Tempo traces, and summarizes the likely root cause
  • A monitoring agent for a media streaming pipeline, or a single creator's live stream, that flags anomalies in real time and annotates the relevant dashboard automatically
  • An incident response agent that queries dashboards and alerts through Grafana IRM workflows to recommend next steps

Requirements

Your project must actively use the Grafana stack at runtime, primarily through the Grafana Cloud MCP server (the official grafana/mcp-grafana server or the hosted Grafana Cloud MCP endpoint). Grafana Cloud AI Observability is a good complement but doesn't satisfy this requirement on its own. The MCP server connection is what's checked.

One important note: the hosted Grafana Cloud MCP server authenticates users interactively, so there's no service-account or machine-token option. Build and record your demo somewhere you can complete the one-time browser authorization. If your project needs a fully unattended, server-side agent, use the open-source Grafana MCP server with a service-account token instead.

Grafana resources

Parallel

Parallel develops web infrastructure for AI agents, with a vertically integrated stack spanning crawling, indexing, search, extraction, reasoning, and monitoring so your agent gets accurate, fresh, and traceable information from the open web.

Ground your agent in live, verifiable information from across the web using Parallel's Search API. See all the details for the Parallel partner track.

Some ideas to spark your thinking:

  • A production research agent that pulls fresh casting, licensing, or market information from the open web to brief a creative team before a pitch
  • A fact-checking agent for media content that verifies claims against multiple live sources and flags anything it can't confirm
  • A trend-spotting agent that monitors the open web for shifts in audience sentiment or emerging topics relevant to a production

Requirements

Your project must actively use Parallel's Search API at runtime, for example through the official parallel-web SDK (Python or TypeScript), a supported integration such as the Vercel AI SDK's @parallel-web/ai-sdk-tools or LangChain's ParallelWebSearchTool, or a Grounding configuration using Parallel Web Search as the search provider. Referencing Parallel in your README alone doesn't satisfy this requirement. The integration must be present in your code.

Parallel resources

ClickHouse

ClickHouse is one of the fastest and most resource-efficient real-time databases and data warehouses, built for data-intensive and large-scale workloads, from interactive, user-facing applications to ad hoc historical analysis. 

Connect your agent to a live ClickHouse Cloud or self-hosted cluster and let it query real data in real time. See all the details for the ClickHouse partner track.

Some ideas to spark your thinking:

  • A chat-with-your-data agent that lets a production team, or a solo creator, ask plain-language questions about viewership, engagement, or performance metrics
  • An agent that monitors streaming or box office data in ClickHouse and surfaces trends as they emerge

Requirements

Your project must actively use ClickHouse at runtime via the official ClickHouse MCP server (mcp-clickhouse), connecting to a ClickHouse Cloud or self-hosted cluster. Use of ClickHouse Agent Skills during development is optional but encouraged.

Note: this must be ClickHouse's own MCP server—connecting to ClickHouse only through Grafana's MCP server does not satisfy this track.

ClickHouse resources

  • Claim $400 in ClickHouse Cloud credits and spin up a service. Heads up: ClickHouse Cloud trial credits are valid for 30 days from signup, which may run out before judging wraps up in October. Plan to extend or renew your instance ahead of the judging period so your project stays live and testable.
  • ClickHouse AI/MCP docs and the official mcp-clickhouse server
  • SQL Playground: public datasets to prototype against before your own data lands
  • AgentHouse demo: a live chat-with-your-data experience backed by ClickHouse MCP (demo is at llm.clickhouse.com, Google login required)

Replit

Replit lets anyone turn an idea into working software using natural language, bringing code, AI, database, authentication, security, testing, and publishing into one platform.

Use Replit Agent to build your project, then deploy it directly on Replit. See all the details for the Replit partner track.

Some ideas to spark your thinking:

  • A crew and call-sheet coordinator that lets a production manager, or a one-person crew, build shooting schedules, assign roles, and get instant alerts when a location or cast member's availability changes.
  • A content approval and sign-off tracker for a post-production team, or a solo editor, that routes cuts, VFX shots, or scripts through reviewers, logs approvals, and flags anything stuck past its deadline.
  • A fan community hub for a creator or small studio, with sign-up, exclusive behind-the-scenes drops, and a voting feature so fans help shape the next episode or release.

Requirements

Your project must be built using Replit Agent as part of the development process, and the finished project must be hosted and deployed directly on Replit (a project URL on a replit.app or replit.dev domain). Projects not deployed on Replit's platform won't meet this requirement, regardless of how the code was written.

Replit resources

Ready to build? Join Agentic Cinema: The Blockbuster Hackathon

Here's how to get started:

  1. Register for the hackathon on Devpost.
  2. Choose your partner. Pick the platform that holds the data or workflows your agent needs to interact with—IBM, Grafana Labs, Parallel, ClickHouse, or Replit.
  3. Build your agent and integrate your partner's product or MCP server to unlock its capabilities.
  4. Start building! Submit your completed project before September 9, 2026 at 2:00 p.m. PDT.

With Google Cloud and Gemini Enterprise as your studio, and five partner platforms giving your agent real capabilities, the possibilities are wide open. We can't wait to see what you build!

Get hackathon news and updates on Discord
Join the Devpost Discord