A second brain
for your browsing

Mnemonic watches what you actually read — not just what you click.
Deep rabbit holes persist. Doom-scrolling fades to nothing.

mnemonic 47 memories · 12 domains
RecentStrongestFading
The Town That Was Buried in Molasses (Twice)
9.5
The Great Molasses Flood of 1919
8.2
Do Octopuses Dream? Scientists Finally Have an Answer
5.8
Can Pigeons Actually Play Ping-Pong?
2.1

History is not a memory

  • A 3am Wikipedia binge and a focused research session look identical
  • A LinkedIn notification you accidentally opened lives next to an article you studied for an hour
  • Good luck finding "that article about the thing" from last Tuesday
  • History is just a list of URLs — no context, no relevance, no way to search by what you actually remember
  • You can't ask it questions. You scroll, hope, and usually give up.

Mnemonic knows the difference

  • Each page is analyzed for type and content — article, docs, tool, social feed — and weighted accordingly
  • 14 minutes reading? Strong memory. One-time lookups fade on their own.
  • Like your real brain — things you return to get stronger, things you ignore disappear
  • Search by keyword across everything you've read — not just URLs
  • Ask "what did I read about vim?" and get actual answers

How it works

01

It reads the room

Mnemonic doesn't track blindly. It analyzes each page — is this a long-form article, documentation, a Reddit thread, or a social feed? It measures your actual engagement: active reading time, scroll depth, idle gaps, revisits. Doom-scrolling Twitter for 20 minutes registers as noise. Spending 20 minutes on a tech blog registers as signal.

02

Memories decay. Or strengthen.

Every page gets a stability score driven by a real spaced-repetition model (the same math behind Anki). Spent 14 minutes on that article? It'll stick around for weeks. Things you keep coming back to build deep roots. One-time lookups disappear on their own.

03

Query your own brain

Search by keyword, or open the AI chat and ask in plain English. "What was that blog post about database indexing?" Mnemonic finds the match and gives the AI full context to answer you. Bring your own API key — Claude, GPT, or Gemini. Your prompts never touch our servers.

Features

Real Memory Model

Based on FSRS (the same algorithm behind Anki). Every page gets a stability score that grows with engagement and decays without it. Documentation pages get a bonus. Quick bounces get penalized.

Forgetting Curve

Memories decay on a real exponential curve, not some arbitrary timer. You can watch a page's retrievability drop from 95% to 30% over days — and see exactly when it'll be forgotten.

PRO

AI Chat

Bring your own API key (Claude, GPT, or Gemini) and talk to your browsing history. "What was that CSS grid tutorial?" The AI searches your memories and finds it.

PRO

Obsidian Vault Sync

Every memory becomes a markdown file with full YAML frontmatter — stability, retrievability, visit count, event timeline. Point it at your vault and your second brain writes itself.

Noise Filter

Gmail, Twitter, Netflix, Slack — all blocked by default. Add your own rules with wildcards and path filters. *.reddit.com blocks everything; reddit.com/r/programming blocks just one sub.

Nothing Leaves Your Machine

Your browsing data stays in Chrome's local storage. Period. We don't run analytics, we don't have a dashboard of your habits, we literally can't see what you browse.

Pricing

Install and get 7 days of Pro free — no card required. The core memory engine stays free forever.

Free

$0

forever

  • Full memory tracking & FSRS model
  • Dashboard with Recent / Strongest / Fading
  • Search across all your memories
  • Custom noise filter rules
  • 30-day memory cap
Add to Chrome — Free Trial

The privacy part

We can't see your data

Not "we promise not to look." We architecturally cannot. Your memories live in Chrome's local storage and never hit a server.

No cloud, no sync, no phoning home

The extension makes zero network requests by default. The only time it talks to our server is to check your subscription status. That's it.

Your key, your AI bill

AI chat calls go directly from your browser to the provider (Anthropic, OpenAI, Google). We never see your API key, your prompts, or the responses.

Our server knows two things about you

Your email address and whether you pay us. That's the entire database row. No browsing data, no usage analytics, no tracking pixels.

FAQ

What exactly gets stored, and where?

For each page you visit long enough (30+ seconds of active engagement), Mnemonic saves the URL, title, a text snippet, word count, and engagement data (active time, visit count, events). All of this lives in Chrome's local storage on your device. Our server only knows your email and subscription status.

What's this "stability" and "retrievability" I keep seeing?

Stability is how long a memory lasts before it fades — measured in days. A 5-minute read on documentation might give you 7 days of stability. Retrievability is the probability you'd still recall the page right now. It starts at ~100% and decays exponentially over time. When it gets low enough, the memory is archived or forgotten.

Why does revisiting a fading page help more than revisiting a fresh one?

That's the spacing effect — the same principle behind flashcard apps like Anki. If you come back to a page when its retrievability is at 50%, the reinforcement bonus is much larger than if you come back when it's still at 95%. The model rewards you for retrieving things that are on the edge of being forgotten.

What's the difference between the 30-day and 365-day memory cap?

Free users have memories capped at 30 days of stability. That means even with heavy engagement and multiple revisits, a memory can't persist beyond about a month. Pro users get a 365-day cap, so deeply reinforced pages can stick around for up to a year.

How does the AI chat actually work? Do you see my prompts?

No. The AI chat runs entirely in your browser. When you ask a question, Mnemonic searches your local memories for matches, packages them as context, and sends the request directly to the AI provider (Anthropic, OpenAI, or Google) using your own API key. We never see the request, the context, or the response.

I cancelled Pro. What happens to my memories?

Nothing gets deleted. Your memories stay exactly where they are. You just lose access to the AI chat, vault sync, and advanced tuning. The stability cap drops back to 30 days, so existing memories above that won't grow further — but they won't be forcibly truncated either. Re-subscribe and everything picks up where you left off.

Can I export my data?

Pro users can sync to a local folder as markdown files, which effectively exports everything continuously. Free users can still see all their data in the extension dashboard. We're working on a one-click JSON export for everyone.

Stop losing your mind

You've already forgotten three articles from this week. Mnemonic wouldn't have.

Add to Chrome — It's Free