RISE, by SyberLabs.
RISE plays text in the browser with timed words, sound and moving graphics. Pilot offer: we set up one excerpt of your text with Stream and Page views and a shareable link, for a fixed quote.
Source on GitHubRISE
A reading app that shows text at a pace you set, with sound and moving visuals.
SAMPLE PASSAGE
Call me Ishmael. Some years ago, having little or no money in my purse, and nothing particular to interest me on shore, I thought I would sail about a little and see the watery part of the world.
Herman Melville, Moby-Dick
Composer in Claude
RISE Composer is a free connector for Claude. Ask Claude for a reading and RISE plays the answer with narration and visuals.
- 01
Compose
The model writes the reading: the words, their timing and the visuals around them.
- 02
Check
RISE checks every choice against what it can play. A reading that fails the check does not play.
- 03
Play
The words appear on a timer, with images, sound and generated motion.
Add it to Claude as a custom connector:
- Name
RISE- URL
https://rise.syberlabs.io/api/mcp
Demo film
RISE demo film
Rooms
RISE’s five main screens, which it calls rooms
Screenshots of a local build of SyberLabs/RISE@998d725 taken on 2026-10-05 in headless Chromium at 1440×900, nothing mocked. The text in every shot is Rural Architecture, from Lyrical Ballads by Wordsworth and Coleridge, a public-domain work served from RISE’s own Library; the fields behind it are RISE’s procedural engines. A silent 25-second screen recording of the same build, Home → Begin → words arriving in Stream (with the few seconds of Preparing session the software renderer needs), is also available. Watch the screen recording (MP4, 25 s, no sound)

01 / HomeHome as it opens: today’s poem, already streaming over its own visual field. 
MenuThe Menu, open on Home: the five rooms, Home, Read, Library, Make and Settings, with Ask for a reading, Guide and Wormhole under More. 
02 / Read · StreamRead, in Stream, ten seconds into a 31-second reading, with the control bar shown: pause, Page view, fullscreen, Visuals, Settings, end. 
02 / Read · PageRead, switched to Page view mid-reading: the same poem set as a page under its edition title, the visual field continuing below. 
03 / LibraryLibrary: curated public-domain works in named editions (the Epic shelf opens with Standard Ebooks translations of The Iliad and Metamorphoses), with Chapel, Rosary, Stations, Journeys, Keystones and Provenance as panes. 
04 / MakeMake, opened from the Menu onto its Workshop pane with Vault, Scriptorium, Visual Lab and Visual Catalog alongside: a new, empty composition; the asset list is the build’s own procedural engines and sourced collections.
Inspect
How RISE checks a model’s suggestion
How RISE checks an AI reading suggestion, using the rules at 998d725: the model makes one request through your connection and can choose only from the supplied options. Two examples show a suggestion being refused without opening a reading.
This map needs JavaScript.
Why
Why combine reading with sound and visuals?
RISE presents a text over time instead of all at once. The reader sets the pace and adds visuals and sound, or switches back to a page view.
The open question is whether this changes how people experience a text. No reader study has been run, so we make no claim that readers prefer it.
Use
What you can do today
Five main areas, with related tools grouped inside each one.
- 01 / Home
Begin today’s reading
Home offers one reading with moving visuals behind it. Press Begin, or choose another reading.
- 02 / Read
Stream or Page
Words arrive over time at a pace you set, or sit on a page you can navigate. Switch at any point; your place is kept.
- 03 / Library
Curated public-domain texts
Browse literature, philosophy, poetry and scripture, with records of each edition, its structure and its rights. The Chapel includes the Douay-Rheims Bible and liturgy with timed pacing.
- 04 / Make
Compose your own
Create readings with text, sound and visuals using Workshop, Scriptorium and Visual Lab. Save them in your browser with the Vault.
- 05 / Ask
A reading from your own model
Describe what you want to read. The model chooses a book and presentation only from the supplied catalog. Connect through OpenRouter, or run Kev-4B locally; model requests use your connection and are not billed to SyberLabs.
Design
How it is built
One reading, two views
RISE divides text into timed units, sets the pacing and prepares one reading session. That same session drives both Stream and Page, so switching views keeps your place.
Book files checked when opened
Book text loads separately from the app, removing 15.4 MB from its JavaScript bundle. Files use SHA-256 content hashes as names, and RISE checks the hash each time it reads a file.
A limit on initial download size
The initial download is about 59 KB across three requests with Brotli compression. Automated build checks enforce a 64 KB limit.
No model runs on the RISE server
The Cloudflare Worker serves the app and its catalog of reading choices. It runs no model and holds no model credential. Six retired model-request routes return HTTP 410 (Gone).
Composer, in ChatGPT
In ChatGPT, RISE lets the model prepare a reading with text, sound and visuals in one tool call. RISE checks the result before making it playable, and the reader presses Begin to start. Rejected readings cannot play.
Automated checks before release
Around 2,800 Vitest unit and integration tests, plus Playwright browser tests, check the app. Releases also check a generated architecture diagram and the built files, with a way to restore an earlier release.
Evidence
- Deployed
The reader, Library, Chapel, Make rooms and the interactive sample run at rise.syberlabs.io.
- Tested
Unit, integration and browser suites run in CI; the first-load budget and the design diagram are checked on every release.
- Measured
Composer in ChatGPT: a controlled developer-mode session on 2026-10-04 produced an admitted Current and narration the reader confirmed hearing. That is partial acceptance on one release, not a public listing.
- Tested
RISE Plus voice, an optional ElevenLabs voice ($8.99 a month): implementation merged; paid launch pending verification. It runs in production on Stripe test mode: on 2026-10-08 a test purchase unlocked it and the voice was heard reading the buyer's own file.
- Not yet
No reader study has been run; release evidence still records zero real-device and stranger-testing records.
Reflects SyberLabs/RISE@998d725 · verified 2026-10-05. Each state is backed by code, a named test, a measurement under stated conditions, or a deployment.
Facts
- Status
- Live · open beta
- Runs in
- Any modern browser at rise.syberlabs.io
- Technology
- JavaScript (ES modules), Vite, Web Audio API, Canvas 2D, IndexedDB, Cloudflare Workers, Vitest, Playwright, GitHub Actions
- AI integration
- Optional and reader-owned: Connect OpenRouter (TypeSafe AI’s hosted Jev), or run RISE locally with a pinned Kev-4B; there is no production Kev deployment, by design.
- Privacy
- Texts you bring and compositions you make stay in your browser. Once you connect your own AI, visual direction sends sections of the reading text to that connection (a text you bring, only with your consent). See the app’s posted Privacy and Terms.
- License
- Apache 2.0 for application code; texts and visual works carry their own terms
- Composer in Claude
- Live as a custom connector: add one named RISE with the URL https://rise.syberlabs.io/api/mcp. Submission to Claude’s connector directory is planned.
- Earlier evidence
- Jev integration in RISE
- Your own model
- Reader-owned AI: Jev and Kev
Other products
- 01 SyberWorkSoftware factory Tests every AI-written code change before a person approves it. Work in progress
- 02 AttestDocumentation testing Runs every step in your docs and marks the ones that fail. Public source · early
- 04 FLYSPACEData canvas for AI assistants Shows which data sources fed each AI answer. Live preview