intelliprompter
Browser teleprompter that checks off talking points as you cover them, scoring each with a parallel Jev Score question.
Overview
IntelliPrompter is a teleprompter of talking points rather than a script. You speak in your own words and any order, and each point is checked off once you have covered it. It is an independent rebuild of a design by Allie at TypeSafe, who built the original for the announcement lists at her stage shows; the README says it cost about $40 an hour on LLMs and at most about $0.50 an hour with Jev.
Every open point is a separate Score question with four levels from no mention to thoroughly discussed, all scored in one request against the transcript so far (the last 6,000 characters). Jev returns a probability-weighted score from 0 to 3; code checks a point off when it reaches a threshold (1.5 by default, adjustable from 0.5 to 2.5). Checked points stay checked and are no longer sent, requests go every 500 ms only when the transcript changes, and confidence is shown but ignored. It uses browser speech recognition and can run with an OpenRouter key on the hosted page or a TypeSafe key locally.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Score | How much has the speaker covered this talking point? | No mention of this topic at all · Topic is mentioned but not expanded upon at all · Topic is discussed · Topic is thoroughly discussed |
How it works
- Browser speech recognition builds a running transcript
- Every 500 ms, open points are scored in one Jev request
- A point is checked off when its score passes the threshold
- Checked points are removed from later requests
Strengths
- README estimates at most about $0.50 an hour versus about $40 on LLMs
- Order-independent: every open point is scored on each call
- Threshold is adjustable live on the page
Caveats
- Relies on browser speech recognition quality
- Hosted page needs an OpenRouter key; TypeSafe key requires running locally
Getting started
Open https://finetuningsingh.github.io/intelliprompter/ and paste an OpenRouter key, or run it locally with a TypeSafe key.At a glance
- Type
- Demo
- Status
- prototype
- Language
- JavaScript
- Runtime
- browser
- Has code
- yes
Use cases and integrations
How curated lists describe it
A teleprompter of talking points that checks each one off as you cover it, using TypeSafe's Jev Score questions.
hellogumbo/awesome-jev
Frequently asked questions
What does IntelliPrompter decide?
How thoroughly each talking point has been covered so far, as a score from 0 to 3.
Why ignore confidence?
It measures how concentrated the probabilities are, not whether a point was covered.
What does it cost to run?
The README estimates at most about $0.50 per hour of speaking with Jev.
Related projects
Cross-platform dictation app where a decision model chooses whether a personal-dictionary replacement actually fits the sentence.
ProjectbetaPythonLiquid Glass voice typing for macOS 26 with an optional Jev fast path that types clean short sentences without waiting for LLM cleanup.
ProjectbetaSwiftMac voice assistant where one Jev call per utterance maps speech to app, volume, Spotify and system actions; an LLM only answers questions.
AppprototypePythonLocal-first meeting transcription and conversation memory app whose live voice path runs its evaluations through Jev.
AppbetaRustToolkit for low-latency voice avatars where Jev can serve as a gate that judges turn ends and turn taking from speech transcripts.
ProjectproductionPythonmacOS workspace for Codex with CodeGraph, Git and terminals, where opt-in Jev history recall finds past sessions and their decisions.
Appbeta