Home

Jewish Learning Suite

AI-powered tools for ancient texts

"For a religious Jew, the highest level of connection to Hashem is spending as much of our Jewish lives learning Torah in as much depth as possible. The task is big, but like many Jews of history, I would 'happily' give up my life in this world instead of giving up on this."

לעילוי נשמת

אברהם חיים בן דוד

L'ilui Nishmas Avraham Chaim ben David

Why I Built This

"For years I was not able to learn in the original Hebrew because I'm dyslexic and I struggle to read and understand Hebrew. So over the years I've been sharpening how I use AI to help me learn as authentically as possible."

The Paradox

Charedi and dyslexic. Learning original Hebrew texts has been a lifelong challenge. Nine years in yeshiva and kollel (2003-2012) without being able to read the texts independently.

The solution: Build AI-powered tools that make authentic Torah learning accessible—not as a shortcut, but as a bridge.

"Learning Directly from the Source—For the last 2 years I have been learning b'Iyun for the first time direct from the source. My chavrusa Itamar is a young Talmid Chacham from JHB and he has been my eyes, reading and translating the words of the Holy Torah, from Gemorah to mefarshim. Our time at night has given me fresh air to breathe."

The Learning Journey

2003-2012

Yeshiva & Kollel

9 years full-time learning

Nightly

Chavrusa with Itamar

B'Iyun learning

Chuburah

Hilchos Brachos

Sun-Thurs 9-10:30pm

Orthodox Jew committed to learning Torah, building relationship with Hashem, and being a kiddush Hashem. Three daily minyanim. Learning every day. The tools exist to extend access, not replace the experience.

The Projects

Talmudic Study App

Real-time AI streaming translation

GitHub →

Watch scholarly translations appear word-by-word as Gemini processes Hebrew/Aramaic source texts. Inline commentary (Mefarshim) with nested exploration. Study journey tracking.

Speed

200 chars/sec

Model

Gemini 2.5 Flash

Layout

70/30 EN/HE

Ohr Avraham Chaim

Professional Hebrew/English book typesetting

GitHub →
"Since I am dyslexic and Charedi, learning original Hebrew texts has been a challenge. That's why I built Ohr Avraham Chaim—it automatically looks up specific Torah text in an open-source DB, then uses Gemini Flash 2.5 Lite to translate to English at $0.1-0.4/million tokens (close to free), then I professionally typeset it into a PDF using LaTeX."

Cost

$0.1-0.4/M tokens

Model

Gemini 2.5 Flash Lite

Output

Print-ready PDF

Source

Sefaria API

Projects:

• Tur & Mefarshim (~50 simanim)
• Nefesh HaChaim (Gate I complete)
• Derech Hashem (in progress)
• Siddur Analysis (858 words, 248 roots)

MCP Sefaria Server

Torah texts through AI

GitHub →

MCP server for querying the Sefaria Jewish text library. Access any text in the Jewish canon through Claude or any LLM.

Talmudic Study App Architecture

Sefaria API → Hebrew/Aramaic Text → Gemini 2.5 Flash → Streaming → React UI
     ↓              ↓                       ↓                  ↓
Commentary    Section Parser        Real-time Chunks    Progressive Display
  Links        (numbered)          (200 chars/sec)      (green highlights)

Key Features:

• Server-Sent Events for streaming
• Translation persistence (Supabase)
• One commentary per section rule
• Nested commentary depth tracking
• User journey analytics
• Rate limiting (Upstash Redis)

Siddur Analysis System

Comprehensive Linguistic Analysis

858

Words analyzed

248

Unique roots

41

JSON files

6.2MB

Structured data

12 fully analyzed prayers (Ashrei, Aleinu, Shema, morning blessings, Torah blessings). 36 Python scripts for linguistic analysis. 6 English translation sources compared. Pure linguistic processing—no AI.

Tech Stack

Talmudic App

Next.js 15, React 19, TypeScript, Supabase, Gemini via OpenRouter, Tailwind 4, shadcn/ui

Ohr Avraham Chaim

XeLaTeX, Python, Sefaria API, OpenRouter/Gemini, babel/polyglossia for Hebrew

Siddur Analysis

Python, JSON, shoresh (root) mapping, cross-text frequency tracking

MCP Sefaria

MCP Protocol, Sefaria API, Node.js

Torah & First Principles

"As a Rabbi with Daas Torah, your profound knowledge and understanding of the Torah's teachings shape your thinking and guide your decisions. Your role involves navigating contemporary issues and moral dilemmas through the lens of Torah principles."

Torah learning and first principles thinking share deep structural similarities. Both require:

Going to the Source

Not relying on summaries. Learning directly from Gemara to mefarshim.

Depth Over Breadth

B'Iyun learning—understanding one sugya deeply rather than many superficially.

Chain of Transmission

Every opinion traced back to its source. No anonymous claims.

Living Tradition

Not dead knowledge—actively interpreted by scholars of every generation.

Who This Is For

Dyslexic Learners

Can't read Hebrew text? AI translates in real-time.

Baalei Teshuva

Starting late? Access classical texts without years of prerequisite study.

Rabbis & Educators

Prepare lessons with AI translation support.

Global Community

Learn Talmud regardless of language background.

The 13-year-old daughter who can speak Hebrew fluently but can't read it due to dyslexia. The adult learner who became frum late in life. The visually impaired student. All deserve access to authentic Torah learning.

The Vision

"Hashem gave the Torah to the whole Jewish people at Mount Sinai and they have an unbroken line of tradition until today. Who else as a people can make a similar claim? There is a rich culture from the world but Torah from Sinai—both written and oral—is the most preserved language model in the world by definition."

The Principles

1. Changing any text is out of the question. Every source must be brought down untouched.

2. Not like other knowledge which is dead—authentic Torah is alive and well in the hearts and minds of the leading rabbis of every generation.

3. Technology as a bridge, not a shortcut. Extending access to those who couldn't read it before.

These tools don't replace traditional learning. They're cognitive prosthetics for those who need them—making authentic Torah study possible for the dyslexic, the visually impaired, the late learner, and everyone in between.

Translation Pipeline

  1. Fetch — Download Hebrew text from Sefaria API
  2. Translate — Use Gemini 2.5 Flash Lite via OpenRouter
  3. Generate — Create LaTeX files with proper formatting
  4. Compile — Build PDFs with XeLaTeX

Prompt: "Output ONLY the translation" — no intro phrases, no explanations.

Torah Content Consumption

Beyond the tools built, the YouTube pipeline tracks thousands of hours of Torah content consumed:

Vayimaen

823 videos

Living Lchaim

781 videos

Ohr Somayach

Q&A with R' Breitowitz

All transcribed, embedded, and searchable through the brain infrastructure. Watch → Transcribe → Embed → Retrieve. Torah content that becomes permanent knowledge.

Talmudic Study App → · Ohr Avraham Chaim → · MCP Sefaria → · All tools →