The narrative knowledge graph for the AI era
We transform films, TV series, manga, and novels into a temporally-aware narrative knowledge graph — so fans can revisit any story with confidence, creators can design with structure, and AI agents can retrieve reliable narrative context.
Demo
A narrated walkthrough of the app: browsing a story's events on a timeline, opening what each moment changed, reading relationships as they shift, and building a query step by step.
Prefer YouTube? Watch the demo there.
Product walkthrough
Real captures from the Knibu iOS build, which runs today in limited access — used internally and by invited testers while we finish the remaining surfaces ahead of a public release.
Nothing in Knibu is a written article. Every screen below is a query: a set of conditions run against the graph and drawn in a chosen form. Change a condition and the same screen answers a different question. Save it and it becomes a card someone else can open and keep pivoting on — that is what a post on Knibu actually is. The nine views below are independent tools, not steps in a flow.
01
Events lay out by when they happened and where. Each one carries the change it caused — who gained what, who moved, what was destroyed — so a plot reads as a structure instead of a summary.
Scrolling one story's event line
02
Detail level is a control, not a fixed property of the record. Step up a level and whole arcs collapse into single entries — a thousand-episode series stays browsable end to end, then opens back up exactly where it matters.
Stepping up a level; arcs collapse
03
Reading a work yourself is one way in. The other is to open a view someone already assembled — its query has already run, so you land on the finished picture rather than a search box. What a view can be built from isn't constrained; this one happens to be a complete timeline drawn around a single object.
Zooming and panning a prepared view
04
Open an event and it resolves into the specific things it moved: identities, relationships, possessions, affiliations — each attributed to the character it belongs to.
Scrolling one event's change list
05
Relationships are typed and time-bound. Drag the timeline and the edges redraw for that moment, instead of collapsing a whole story into one flat summary.
Dragging the timeline; edges redraw
06
A hierarchy is data too: which posts exist, who holds each one, and when it changed hands. Move to a different era and the chart repopulates with whoever held those posts then.
Changing era; post-holders swap
07
Put characters side by side against the same axis and their affiliations, abilities and status can be compared decade by decade rather than read one biography at a time.
Stepping across four character tracks
08
Filter by language, genre, theme, release window and box office. Results come back as a graph you can keep pivoting on, not a list that ends the conversation.
Adjusting filters; results respond
09
Bind a numeric attribute — here, box office — to node size, and the shape of a whole result set becomes readable at a glance.
Results rescaling by box office
Why now
When a sequel arrives years later, audiences struggle to recall what happened. The barrier to re-entry hurts engagement and box-office returns.
Maintaining continuity, character arcs, and timeline logic across a large story is hard — and getting harder as stories grow.
General models hallucinate plot details; web search returns noisy, inconsistent results. There's no trustworthy, structured source of narrative truth.
The product
The same temporally-aware engine, surfaced for three audiences.
For fans
Catch up on the stories you love, before the next chapter drops.
For creators
Build your story as a structure, not just a document.
For developers & AI agents
Reliable, time-aware narrative context for AI agents.
How it works
01 / Ingest
We process story content into structured narrative data: events, entities, relationships, and their changes over time.
02 / Structure
Information is organized into a temporally-aware knowledge graph that preserves internal logic and consistency.
03 / Access
Users query in natural language; the system retrieves the right facts and presents them through clear, purpose-built views — or serves them to AI agents via API.
The complexity lives in the graph; the experience stays simple.
Where we are
Built for scale
Knibu runs an AI-intensive, cloud-native stack. Large language models do the work of turning narrative information into structured, time-aware graph data, and the graph itself is hosted on managed cloud infrastructure. Our compute curve is driven by what comes next: opening the app to the public, serving real-time queries to fans, creators, and AI agents, and running AI-driven natural-language query across the graph. Cloud compute and managed AI services are central to that.
Team
A one-person company building the full Knibu stack, with 95% of the code written through AI-assisted development. Background in engineering, coding, and product management; graduate of the University of Wisconsin–Madison; four years of experience as a founder.
We're opening early access to fans, creators, and developers. Tell us how you'd use Knibu.
We'll reply from contact@knibu.com
