For Investors
Teaching reading comprehension, one word in context at a time.
Symporead is an AI-assisted reading environment, built on second-language acquisition research rather than gamification — for learners and schools who want depth, not streaks.
The problem
Reading breaks the moment understanding does.
A reader who stops to look up a word loses the sentence, the paragraph, and often the will to keep going. Dictionaries define words in isolation; they can't tell you what a word means here, in this sentence, in this book. Generic AI chat tools weren't built for reading — they answer questions, they don't sustain a reading habit.
The approach
Meaning derived in context, never out of it.
Every page a reader opens is tokenized into sentences and words. A click sends the word and its sentence to an LLM, which returns the in-context meaning immediately — then indexes that sentence so the reader can later search every context a word has appeared in. Vocabulary is built the way it's actually acquired: in use.
Phase 1 — built
A working product, not a deck.
Context-Aware Vocabulary Engine
Click any word and an LLM derives its precise in-context meaning — a real replacement for dictionary lookups, which strip words of the sentence that gives them meaning.
SympoChat — Realtime Socratic Dialogue
Chapter-level discussion that checks comprehension the way a good seminar leader would: with questions, not quizzes.
Omnipresent Voice Input
Reading and discussion without breaking flow to type — voice is a first-class input everywhere in the product.
Teacher Dashboard
Production-ready reporting so an instructor can see comprehension and vocabulary growth across a class, per chapter.
Phase 2 — next
SympoChat 2.0
A system-initiated conversational engine that proactively scaffolds a reader through comprehension gaps in real time, grounded in a secure knowledge graph of curricula.
Live comprehension signals
The AI notices a disconnection as it happens and steps in with a scaffolding question — not after a chapter ends.
Proprietary curricula, ingested
A school's own curriculum becomes a secure knowledge graph the dialogue engine can draw on for context injections.
Rigorous, academic tone
Advanced RAG and multi-agent orchestration in service of a Socratic dialogue that reads like a seminar, not a chatbot.
Why this works
Grounded in acquisition research, not trends.
The Input Hypothesis
Krashen, 1985
Comprehensible input — language just beyond a learner's current level, understood in context — drives acquisition. It's the theoretical basis for reading with in-context help instead of rote memorization.
Creativity, Competition & Partial Productivity
Goldberg, 2019
Word meaning is learned as partially abstracted patterns shaped by context and repetition — the reason in-context definitions generalize better than dictionary entries.
Finding Structure in Time
Elman, 1990
Foundational work on representing sequence and context computationally — the lineage behind using language models to track meaning across a sentence.
Team
Schools, engineering, and AI — covered.
Rae — CEO
5+ years of educational consulting for over 100 elite US boarding schools.
Ezio — CTO
10 years building web products; leads the engineering behind the reading and vocabulary engine.
Katherine — Chief AI Officer
Former PayPal and Tier-1 Web3; 5+ years in data science, AI, and product strategy.