everything you need to know about MYNAH. something missing? send us a question.
MYNAH is a personal knowledge instrument. it passively ingests the long-form content you read and listen to — articles, essays, podcasts — and lets you converse with an AI that has read everything you've read.
ask it a question, and it searches your reading history, surfaces the relevant passages, and gives you a direct, sourced answer. not the open web — just your corpus.
the name comes from the mynah birds in Aldous Huxley's Island, trained to call out "attention — here and now." it's a small reminder to be present.
the product philosophy is the same: technology that pushes you toward life, not away from it. no algorithmic feed. no engagement metrics. no notifications designed to pull you back. MYNAH is a tool you use, not a platform that uses you.
people who read a lot and want to remember more of it. people who finish an article, think "that was interesting," and two months later can't remember where they read it or what it said.
it's not for power users who want to manage a second brain manually. it's for people who want their reading to compound — quietly, without extra friction.
a few meaningful differences:
anything long-form and text-based. MYNAH is designed for content worth reading:
pages under ~300 words are rejected as too short. MYNAH is for long-form reading, not tweet-length content.
install the extension, sign in. when you're on an article you want to save, click the MYNAH extension and hit SAVE TO BRAIN.
the extension extracts the article body (title, author, full text), strips the noise, and sends it to your library. it's chunked and embedded in the background — usually done before you've finished reading.
MYNAH gives you a phone number when you sign up. text any URL to that number and it goes straight to your library — no app required.
this is useful for apps with non-standard share sheets, or when you're on mobile and just want to fire a link off quickly.
on iOS, MYNAH monitors what's playing in the background. when it detects a podcast, it looks up the episode metadata and saves it to your library automatically.
you can toggle this behavior in settings. it checks every ~15 minutes in the background, and also runs when you open the app.
type a question in the app. MYNAH embeds your question as a vector, runs a similarity search across all your ingested chunks, and finds the most relevant passages from your reading history.
the AI then synthesizes an answer grounded in those passages — and shows you which articles it drew from. if it can't find relevant material, it says so.
MYNAH uses Claude (by Anthropic) for generating answers, and Voyage AI for creating the vector embeddings that power the search.
the model is instructed to be direct and opinionated — it will tell you if your sources point clearly in one direction, rather than hedging every answer.
no. MYNAH's answers are grounded only in your reading history. if you haven't read anything on a topic, it will tell you: "I don't have enough reading history on that topic yet."
this is intentional. the value of MYNAH is that you know exactly where every answer came from — your own corpus, your own judgment in saving those articles.
yes. every article you save adds context. the more you read on a topic, the richer and more nuanced MYNAH's answers become on that topic.
your corpus is also personal — MYNAH reflects your specific reading, not an average of what everyone has read. over time it becomes a genuinely individual instrument.
yes. your library is private by default and scoped entirely to your account. no one else can see what you've saved or ask your corpus questions.
MYNAH doesn't build a public profile, doesn't aggregate reading data across users for recommendations, and doesn't have an algorithmic feed. your reading is yours.
yes. you can toggle incognito mode in settings. while incognito is on, nothing is saved to your library — the extension and share sheet are effectively paused.
article content is sent to Voyage AI for embedding (anonymized chunks, no PII) and to Anthropic's Claude API when you ask a recall question. both are standard API calls with no persistent data storage on their end beyond what their standard terms allow.
your library and embeddings are stored in your personal Supabase database, isolated by your user ID with row-level security.
at minimum: install the Chrome extension and sign in. that's it — start saving articles.
for the full experience, also download the iOS app. it gives you recall on the go, podcast detection, voice notes, and the share sheet for saving from any app.
iOS 17 or later. the app is built with SwiftUI and SwiftData.
not yet. the current surfaces are the Chrome extension (capture) and the iOS app (recall + library). a web console is on the roadmap.
MYNAH is in early access. pricing hasn't been announced yet. reach out if you want early access.
send us a note and we'll get back to you.