ArcMotivate - Creative Storyteller for your future

"What's your dream career? - Explore new stands of your future by making use of Gemini's multimodal generation. Originally crafted for audiences 8-18, this can also be used for any person.

Comment

ArcMotivate -- Creative Storyteller for Your Future

Inspiration

"What is your dream?"

It's a question many children are asked when they are young. But imagining a "dream career" can be difficult when you don't know what's available.

Traditional career guidance is often static and informational, while young people tend to explore ideas through stories, play, and conversation. Career directions are often influenced by parents or peers.

We wanted to create a more engaging way for young people to explore real careers --- one that begins with their curiosity and turns it into a personalized journey of discovery.

ArcMotivate transforms career exploration into an interactive narrative experience that helps young people reflect on their interests, discover new possibilities, and imagine what their future could look like.

What it does

ArcMotivate is an interactive career exploration tool that uses multimodal AI to help young people discover potential career paths based on their interests, hobbies, and conversations.

It can also be used by anyone to optimise their future career trajectory.

Conversational Exploration

Users interact with a responsive chat interface that encourages reflection about interests, skills, and aspirations. The system dynamically adapts to their responses.

Interactive Career Discovery

During the conversation, ArcMotivate surfaces Career Tiles---interactive cards containing:

  • Job titles
  • Industry skill tags
  • Exploration prompts
  • Safe Google search links

These tiles help users connect their interests to real-world careers.

Personalized Storybook

At the end of each exploration session, ArcMotivate generates a visual story summarizing the user's journey:

  • Hero Recap highlighting key interests
  • 3‑Panel Identity Comic representing a narrative arc (Spark → Experiment → Direction)
  • Custom Avatar
  • "Postcard from Future You" visualizing a potential future scenario

Users can replay the experience and explore different career paths each time.

How we built it

ArcMotivate was built using a combination of Antigravity Rapid Prototyping and manual fine tuning of the system to allow the project to be delivered in a short amount of time.

The project is deployed exclusively on Google Cloud Run as it provides the best developer experience using Gemini Models and utilises Github Build on push to speed up management of revisions.

AI Models

We used the Google GenAI SDK with a combination of:

  • gemini-3.1-flash-lite-preview
  • gemini-3-flash-preview

These models power:

  • Structured narrative generation
  • Prompt generation
  • Multimodal outputs

Structured Outputs

All model responses follow strict JSON schemas, ensuring the system receives predictable structured outputs for:

  • narrative text
  • image prompts
  • career suggestions
  • resource links

Parallel Processing

Generating multiple personalized images sequentially caused latency. To keep the experience responsive, we implemented concurrent generation using:

concurrent.futures.ThreadPoolExecutor

This allows the avatar, comic panels, and future postcard to render simultaneously in the background.

Challenges we ran into

Image Generation Latency

Generating several images sequentially slowed the experience. Parallel generation significantly reduced waiting time and preserved a smooth interaction.

Hallucinated Links

Language models frequently generate fabricated URLs. We implemented defensive prompting and fallback logic so all suggested resources resolve to safe Google Search links.

Visual and Cultural Consistency

Text‑to‑image systems often default to generic or US‑centric aesthetics. To address this we implemented a visual style guide enforcing:

  • A grounded futuristic pixel‑art style
  • A UK‑appropriate cultural context

Accomplishments that we're proud of

Narrative‑to‑Visual Pipeline

We built a pipeline where a single LLM call analyzes the conversation, produces a structured narrative arc, and generates prompts for a cohesive 3‑panel comic.

Robust Demo Architecture

If image generation fails or an API call times out, the system falls back to default assets and narrative summaries so the experience never breaks during demos.

Grounded Personalization

All generated narratives and visuals are grounded in details extracted from the user's conversation, ensuring outputs feel relevant and personal.

What we learned

Defensive Prompting

Separating system instructions from user signals was critical to prevent prompt contamination when chaining outputs into image generation.

Multimodal AI Orchestration

We learned how to chain structured text generation, prompt generation, and image generation while maintaining a consistent narrative across outputs.

What's next for ArcMotivate

Career Progression Knowledge Graphs

We plan to introduce an interactive graph that connects:

  • a user's hobbies
  • beginner skills
  • education pathways
  • related careers

Expanded Visual Styles

Future versions will allow users to choose different visual styles for their storybook (watercolor, comic, or 3D render).

Deeper Career Data Integration

We aim to integrate richer datasets to provide clearer pathways from interests to real-world career opportunities.

Built With

  • buckets
  • gemini
  • genaisdk
  • google-cloud-run
  • google-secret-manager
  • gradio
  • html5
  • nano-bana-2
  • python
Share this project:

Updates