Interactive reading
BOOKBLOOM
Literature that talks back.
BookBloom is an interactive reading platform powered by the ERES format — a structured standard for AI-native literature. Characters respond to questions in context. Scenes generate from page metadata. Ambient audio scores the atmosphere chapter by chapter.
ERES-powered reading
Open the bundled ERES title to experience character chat, scene visualization, and ambient audio — all from a single reading view.
Talk to the characters
Open the character panel at any point in the story and ask questions directly. Responses are grounded in the narrative, not pulled from the open web.
Visualize as you read
Request scene generation per chapter. BookBloom renders imagery from metadata embedded in each page — no generic stock art involved.
Hear the world
Ambient audio scoring driven by scene tags evolves with each chapter, building atmosphere that matches the emotional weight of the text.
Why BookBloom
Reading as a two-way experience
Traditional e-books are static. BookBloom is a bet that the next generation of literature should be structured for AI interaction from the start — not retrofitted after the fact.
The ERES format embeds structured metadata directly into the book — character knowledge graphs, scene descriptors, and audio tags — so every interactive feature is grounded in the author's intent rather than generated from scratch.
The result: you can ask a character a question and get an answer that respects what they would plausibly know at that point in the story. You can request a scene image and get something that reflects what the text actually describes, not a generic landscape.
Character knowledge scoping
Character responses are scoped to the current story position — the AI cannot reveal what a character has not yet learned.
Reading journey
Five moments from first page to full immersion
The prototype demonstrates five interactive layers — each triggered naturally as you move through the story.
01
Open the title
Load an ERES-format book. A bundled title with full character, scene, and audio metadata is included — no downloads or setup required.
02
Read and navigate
Move through chapters normally. Page-level state tracking keeps your position and context synchronized across all three interactive systems.
03
Talk to the cast
Open the character panel and ask any character a question in context. The AI responds using knowledge embedded in the ERES metadata — grounded in the story, not the open web.
04
Visualize the scene
Request scene generation from the current chapter. The platform uses page metadata to direct image generation — grounded in what the text actually describes.
05
Hear the atmosphere
Ambient scoring activates from the scene tags embedded in the chapter. The audio builds alongside the narrative instead of playing as a static background track.
Feature deep dive
Character chat, scene generation, ERES format, and platform architecture
Four capabilities that define what makes BookBloom different from a standard e-reader.
AI character conversations
A frosted-glass chat panel surfaces inside the reading view. Each character's knowledge is scoped to what they would plausibly know at the current point in the story.
- Context-aware responses: character knowledge does not exceed story position.
- Inline panel design — reading view stays primary, chat is secondary.
- Character profiles and knowledge graphs embedded in ERES metadata.
Scene visualization
Each chapter carries scene metadata that describes setting, lighting, mood, and key objects. The architecture uses this to direct image generation rather than interpreting free text.
- Scene generation grounded in structured ERES page metadata.
- Chapter-level granularity — distinct visuals per chapter, not per book.
- Chapter-level granularity — distinct visuals per chapter, not per book.
ERES format
Enhanced Reading Experience Standard is the proprietary format BookBloom reads. It extends plain text with structured metadata layers for AI interaction, scene generation, and audio scoring.
- Page-level state tracking with embedded entity tagging.
- Character knowledge graphs and scene descriptors per chapter.
- DRM layer: AI model access authenticated against verified ERES content.
Platform architecture
BookBloom runs a scalable cloud infrastructure with dedicated model inference, a responsive reading interface, and ScaiCo-powered multi-agent sessions for character conversations.
- Cloud-native backend with session and character state management.
- Responsive reading interface rendering ERES content with inline AI panels.
- Scalable infrastructure for parallel model inference workloads.
Product overview
Capabilities, roadmap, and resources
What BookBloom does today, what's coming next, and where to learn more.
Core capabilities
- Bundled ERES title included — open and start reading immediately.
- AI character conversations with context-aware responses scoped to story position.
- Chapter navigation with page-level state tracking across all interactive layers.
- ERES format reader — structured metadata for AI interaction, scene rendering, and audio.
- Cloud-native infrastructure with scalable model inference.
What's next
- Expanded scene visualization styles
- Advanced ambient audio layering
- Third-party ERES title marketplace
- ERES authoring tools for publishers
Explore further
- Open the bundled ERES title and start reading.
- Read character responses scoped to the current story position.
- Review the ERES format specification and metadata layers.
- Explore the platform architecture overview.
Powered by ScaiCo
Multi-agent AI, character sessions, and scalable inference
BookBloom runs ScaiCo as its multi-agent infrastructure. Character conversation sessions, credit-based turn tracking, and model orchestration all go through the ScaiCo stack.
ScaiCo powers character sessions
ScaiCo orchestrates multi-agent character conversations, manages credit-based turn billing, and maps each exchange to the story position stored in the ERES session state.
- Character conversation sessions launch from the reading view — each powered by ScaiCo with per-session credit tracking and character-specific agent configurations.
- ScaiCo enforces the story-position scope: agents are given only the character knowledge that is valid at the reader's current chapter, preventing spoiler drift across sessions.
- Scene generation and audio scoring will route through the same ScaiCo infrastructure, enabling unified credit accounting across all three interactive layers in one session.
What is ScaiCo?
ScaiCo is Swaibian's multi-agent AI orchestration platform. In BookBloom, it powers the character conversation system — coordinating specialized agents for in-context dialogue, narrative scoping, and credit-based turn management across the reading session.
Next step
Open a story and start a conversation
A bundled ERES title is included. No sign-up required — open the book, talk to the characters, and see scene visualization in action.