The Bookmark Graveyard Is Worth Solving

01 /The Concept

Zenfetch positions itself as a tool for "acquiring digital wisdom"

Wisdom accumulates through knowledge and experience. Experience, for now, is outside the product's scope.

But knowledge – that we gather from the internet every day. Deliberately searching, accidentally stumbling across useful articles, guides, videos, documents. We save them to phone notes, browser bookmarks, or pinned tabs we plan to read later.

The problem: most of that saved material just sits there. We forget what we've saved, where we saved it, or what made it useful. Scrolling through hundreds of bookmarks is tedious. And the more we accumulate, the more intimidating it becomes to look through any of it.

Zenfetch is built to fix that.

The core product is a browser extension. When you land on a page worth keeping, you click the extension and its contents are saved to your Zenfetch database. This alone isn't novel – there have been dozens of bookmark-saving extensions.

What's different is what you can do with the content afterward.

You don't search Zenfetch – you ask it. "What have I saved recently about topic X?" returns a list of relevant links and documents. "What have I saved that's similar to this article?" returns related material. If something is too long to read in full, ask for a summary – of an article, or even a video.

The saved database isn't a list of links. It's a training corpus for a personal AI librarian, who manages your accumulated reading and is available to chat. You can ask questions in natural language; the assistant answers using your own saved content, surfacing things you bookmarked and forgot – or never finished reading.

The assistant can also draft original documents on a given topic, drawing primarily on your saved material rather than on generic internet knowledge. If you want a piece on a specific subject written from a particular angle, you first save the sources that reflect your perspective, then ask Zenfetch to write – and the output reflects your curated viewpoint, not a generalized summary.

Supported content types: websites, articles, social media posts, forum threads, Google Docs, PDFs, and YouTube videos.

Price: $14.99/month.

Zenfetch went through Y Combinator in winter 2022, receiving the standard $500K at the time. But the current product only went live 19 hours before this review was written.

02 /Why It Matters

Was Zenfetch quietly building this product for two years? Not quite.

They came out of Y Combinator with a completely different product – a sales tool that used AI to score leads based on engagement history. Then came a pivot, and now they're launching something entirely new.

Looking across Y Combinator alumni, this is common: a large share of companies that get into the program change their core idea, sometimes radically, even when the original idea was compelling enough to earn a spot. That's healthy. Y Combinator drills into founders that the only real test of an idea is whether actual users want it – and that the ability to discard ideas that fail that test without attachment is a core founder skill.

On the current Zenfetch idea itself: the problem is real. We save far more than we use, partly because retrieval is too hard and partly because we forget what we've stored.

Rewind tackled a version of this by recording everything that happens on a user's computer – screen and audio, including meetings – creating a searchable memory of everything the person has done and said. Rewind raised $16.4M in November 2022 and another $17.9M in July 2023.

But beyond the specific startups, this is a genuinely large potential market. We all use some form of digital notes or bookmarks. The question is whether the right product can unlock that latent demand.

Evernote came closest for a while. From 2008 to late 2011 it grew from zero to 13.6 million users, adding 40,000 new users per day at its peak. By mid-2016, it had 200 million registered accounts. Then things went sideways. But the market potential Evernote demonstrated was unmistakable.

03 /Opportunities

One path is to chase the early Evernote trajectory on the notes and bookmarks market – but with the capabilities that AI now makes possible. Zenfetch and Rewind are among the current experiments worth studying, iterating on, or going beyond.

A separate direction grows out of the Zenfetch draft-generation use case.

This is an early version of a personalized AI that reflects your own knowledge and perspective. Personal AI, [covered in early 2023](/review/vzorvat-rynok-obrazovanija), raised $13.7M for exactly this concept – train an AI on your own writing, and it can handle messages and responses in your voice without requiring your time.

But the concept can be generalized further. You might not want a digital twin of yourself. You might want an AI that reflects a specific set of views – for the purpose of debate, stress-testing an argument, or exploring a perspective you don't already hold.

Dopple, [covered in late 2023](/review/individualnost-cepljaet-i-prinosit), is building a marketplace of AI chatbots modeled on historical figures and fictional characters.

And there's an extension of Zenfetch's own model worth imagining: what if saved content could be organized into separate folders – and the AI could write or analyze using the contents of a selected folder? One folder per author, one per topic with multiple viewpoints, one per project. That would turn Zenfetch from a personal archive into something more dynamic and compositional.

There's still a lot of unexplored territory at the intersection of personal notes, curated knowledge, and AI personas. What would actually be useful for you?