The Next Chapter of Readwise: Our Own Reading App

by Daniel Doyon and Tristan Homsi

Our mission here at Readwise is to improve the practice of reading through software by an order of magnitude. This sounded just as ambitious when we originally wrote it in 2018, but we’ve made some meaningful steps since: thousands of customers, hundreds of millions of highlights, and more.

Today, however, we're excited to announce a huge leap: We've built our own, fully-integrated reading app.

Currently, our reading app (which we refer to as "Reader" for the time being) is a more powerful, more flexible version of the classic read-it-later app. If you've used Instapaper or Pocket, it's like those except it's built for 2021 and beyond. This means:

  • It’s made to handle modern and established content alike, from Twitter threads to PDFs, and everything in-between.

  • It serves both casual and power users, with the flexibility to accommodate a variety of consumer, professional, and academic use cases.

  • It’s designed with a local-first, cross-platform architecture enabling blazingly fast interactions and full-text search across all of your devices (even offline).

  • It connects seamlessly with all your other tools for thought such as Roam Research, Notion, Evernote, and Obsidian.

  • It embeds powerful workflows to help you conquer content overload — one of the most acute pain points of read-it-later power users.

  • It's being actively developed with responsive customer support and rapid feedback loops informed by beta testers who are already reading in the app for hours a day.

There’s so much more to the reading app than this, but we didn't want to bury the lede.

We're now in position to reimagine aspects of the digital reading experience itself, from how you annotate a document, to how you navigate. Readwise as you know it today isn't going anywhere, but this is our future.

If you’re ready to dive deeper, this post explains:

  • Why we've decided to build our own reading app;
  • What the reading app actually does; and,
  • How we're going about product development.

As the expression goes, let's start with why.

Why are we building our own reading app?

The simplest explanation for why we're building our own reading app is this is what our users have been asking for!

The chief complaints we receive these days are about the limitations of the apps upstream from Readwise and, as a layer built on top, we’re not in a position to solve those problems. For example:

  • Why can’t I easily return to the context surrounding a specific highlight?
  • Why can’t I highlight and save images, tables, and rich text formatting, despite these being crucial to the reading experience?
  • Why does it take so long for my highlights to sync?
  • Why is it so hard to get highlights out of my PDFs?

There are dozens of other complaints like these and we’re done telling our users, "Tough luck. There's nothing more we can do." Our only option at this point is to craft our own reading experience intended for our high expectation customers.

But this isn't the whole story. Addressing the complaints above only scratches the surface of what should be possible with software-augmented reading.

Our mission, revisited

Let us be clear on what we mean when we use the term "reading" throughout this post.

When people categorize books, they typically divide them into either fiction and nonfiction. This kind of works, but what about blog posts? Newsletters? Long-form journalism? Twitter threads? Journal articles?

Fiction versus nonfiction now seems outdated so we've adopted an alternative dichotomy: "reading for entertainment" versus "reading for betterment".

Reading for entertainment is any reading where the primary pursuit is pleasure. This includes reading fictional books, of course, but also most newspapers, magazines, and casual web browsing. This kind of reading actually has been disrupted by technology, most notably by social media which is, at its core, a form of entertainment.

Reading for betterment, on the other hand, is where we see the opportunity for software to improve reading by an order of magnitude.

Reading for betterment can be pleasurable, but pleasure is not the ultimate intent. Instead, there is a deeper purpose. This might be to expand the boundaries of knowledge through original research; to become a more eloquent speaker; to inquire into the meaning of life; or to simply internalize a memo for work.

Helping people to better achieve the objectives that inspire them to read is what defines our mission because, if successful, it would unlock untold amounts of human capital.

With the right software, we believe this is possible.

Why now?

We’re certainly not the first to lament the ineffectiveness of reading compared to what it could be with technology. Perhaps the most famous call to arms was Vannevar Bush’s As We May Think, published in 1945, in which he imagined a memory expanding device called the memex. There have been several such futurists since: from Douglas Engelbart and Ted Nelson at the dawn of personal computing, to Alvin Toffler and Stewart Brand at the dawn of the internet, to Kevin Kelly and Andy Matuschak today.

This begs the arguably more important question of not why but why now?[1]

As we wrote in 2018, technology has successfully disrupted traditional publishing and distribution models, but the experience of reading is yet unchanged. Software has merely moved the words from a printed page to a backlit screen leaving us to wonder:

  • Why is it still so hard to search our books, articles, notes, and highlights? Why are we using Google to search the entire internet and find something we read last week?

  • Why is it still so much harder to navigate a digital book than paper? Why can’t we easily save a place, jump backwards, hop forwards, and cross-reference tables, charts, and other other documents side-by-side as nimbly as we can with paper?

  • Why is reading a long-form document on our phones, tablets, or desktops still so distracting? Why doesn’t software help us transition into a relaxed, concentrated flow state?

  • Why are we flooded with a deluge of books, articles, and papers? Why doesn’t software help us decide what to read now, what to read later, and what to read never?

  • Why do we still spend 10 hours reading a book only to forget nearly everything we understood in the moment? Why doesn't software help us internalize those things we want to apply to our personal and work lives?

  • Why do we still need to stitch together 7 different apps to unify our digital reading? Why do we need an ebook app for nonfiction, a read-it-later app for long-form web articles, an email inbox for newsletters, an RSS reader for feeds, a PDF reader for memos and journal articles?

We believe these unanswered questions explain why, after more than fifteen years, ereading has struggled to put a dent in the nonfiction book market.

"Publishers of books in all formats made almost $26 billion in revenue last year in the U.S., with print making up $22.6 billion and e-books taking $2.04 billion, according to the Association of American Publishers’ annual report 2019. Those figures include trade and educational books, as well as fiction." (Lucy Handley, Physical Books Still Outsell E-Books — And Here's Why)

Said differently, ebooks are just not yet better enough, at least for nonfiction readers, to justify switching from physical books.

This is not because the practice of reading has somehow been perfected in physical form and books are among the few domains insulated from the insatiable appetite of software eating the world. It’s because the technology industry thus far has focused on building out hardware and distribution infrastructure rather than reimagining what a reading application could be.

At first, we lacked the right hardware. When the personal computer originally came out, it quickly disrupted the practice of writing. Even the earliest word processing software offered massive advantages to writing by hand or by typewriter. But reading on a personal computer was markedly worse than reading printouts or paper books, magazines, and newspapers. Imagine reading an entire book on this:

The landscape of hardware began to change approximately 15 years ago as a new class of computing device intended for consumption rather than productivity emerged: the smartphone and its sibling, the tablet.

This set off a brief period during the late 2000s when it seemed like reading technology was set to flourish. During this time we saw the advent of Kindle, Pocket, Instapaper, Google Reader, Goodreads, and others. But looking back, it’s now obvious that this era was truly about controlling the distribution of content in the new mobile-first world rather than advancing the practice of reading itself.

Jeff Bezos famously frustrated the special ops team responsible for Fiona — the internal codename for what we now know as Kindle — by demanding that the first generation device include cellular connectivity.

The [designers] couldn't understand how the economics of the wireless connection could work. "But you can't do that," [the chief designer pushed back].

"I'll decide what I can do," Bezos said. "I'll figure this out and it is not going to be a business model you understand. You are the designers, I want you to design this and I'll think about the business model." (Brad Stone, The Everything Store)

It's safe to say that Bezos wasn't so focused on Whispersync because he thought this would somehow improve the practice of reading. Instead, he knew the connectivity would be key to extending Amazon’s distribution advantage to the digital world.

But now that the hardware and distribution layers are set, the infrastructure is in place to innovate the reading experience on the application level.

What is this better reading app?

If you're thinking, "All this sounds cool and everything, but it's a little handwavy," we agree. Let us actually describe what our reading app does (or will do soon) in tangible, product terms.

One framework we've found especially helpful to structure such product discussions is to separate reading into three components: before, during, and after you read.

If we can build software that helps you be twice as effective in each of these stages — choose twice as efficiently before you read, comprehend twice as deeply while you read, and retain twice as much after you read — the compounded effect will be the Thielian 10x improvement we’re shooting for. [2]

After Reading: Beginning with the end

We actually began Readwise by focusing on what happens after you read. To us, this means addressing the problem of retention. Starting with this problem, we believed, would make the greatest immediate impact on reading for two reasons.

First, from a practical perspective, it’s a common pain point that you quickly forget most of what you read. Meanwhile, there were no other products offering a purpose-built solution to retaining more of what you read.[3]

Second, from a theoretical perspective, the best way to start growing more in any domain is to first lose less. We see this first principle everywhere:

  • In a subscription business, you're better off reducing churn before increasing sign-ups.
  • In personal finance, you're better off spending less before earning more.
  • In investing, you're better off stopping losses before seeking gains.

If you want to fill a leaky bucket, you first need to stop the leak. If you want to get more out of reading, you first need to forget less.

Readwise, as our users know it today, addresses the problem of retention through two value propositions. First, through a resurfacing feature that makes it easy to consistently review your notes and highlights from existing reading apps such as Amazon Kindle, Apple Books, Google Play Books, Instapaper, Pocket, Medium, PDFs, Twitter, and even from physical books using OCR. Second, through a syncing feature that makes it easy to continuously and automatically synchronize all your highlights to your note-taking tool of choice such as Evernote, Notion, Roam Research, and Obsidian.

An in-depth discussion of resurfacing and syncing is beyond the scope of this post since you can just try the existing Readwise product for yourself. One point we do want to make though is that for the typical Readwise user, we've already hit the 2x target improvement on retention (the after phase of reading).

How do we know this?

Qualitatively, our users tell us that once they get in the habit of using Readwise, they can no longer read paper books. It's too frustrating to invest so much time into reading a book only to forget its key points a few weeks later. (This is our favorite feedback by far!)

Quantitatively, we've been sending out Superhuman's now infamous product-market fit survey for over three years. In addition to the trademark question of “How disappointed would you be if you could no longer use Readwise?”[4], we also ask "How much more would you say you retain from your reading with Readwise than without?"

Across thousands of surveys, the average user reports retaining 2x more. We know this isn't perfect science, but it’s something.

With the new Readwise reading app, not only will these resurfacing and syncing features not go away, they will be enhanced through tight integration into the reading experience. Some examples:

  • Whenever you take a new highlight, you can see that highlight instantly propagated throughout your note-taking apps.
  • Whenever you annotate a document, you can forge connections to other documents, authors, notes, and highlights — using the same fuzzy search, autocomplete and backlinking you'd expect in any modern tool for thought.
  • Whenever you tag or add an annotation to a resurfaced highlight, you can see that update reflected in the margin of the original document (and vice versa).

In addition to expanding the quantity of things you can retain, the Readwise reading app also facilitates the ease with which you can recall. This brings us to blazingly fast, full-text search.

Interlude: Full-text search

A key feature of Vannevar Bush's memex — imagined 76 years ago — was easy, instantaneous search of everything you've ever read. Why we can't do this already, we have no idea. It’s not like some yet-to-be-discovered technological breakthrough is required. The requisite technology has existed for decades! Yet for some reason, most of us are still forced to google the whole internet to find something in our own library we read last week. This makes no sense.

To be fair, there are some power users who've kludged together a system that gets all their content into a single note-taking app where it can then be indexed and searched. Readwise is even a key component in most of these pipelines. But the labor involved in setting up and maintaining a system like this is painstakingly prohibitive for 99% of the population.

Sources: Eva Keiffenheim & Ravi Kurani

The Readwise reading app is built on a local-first foundation enabling blazingly fast, full-text search of your entire library of books, articles, annotations, and highlights. You can find whatever you're looking for, even if all you can remember is a single word. Even offline.

Before Reading: Conquering content overload

If improving what happens after you read is about fixing forgetting, improving what happens before you read is about conquering content overload. There's one pain point we hear more than any other from our users and it’s that they’re drowning in too much content with not enough time. But before we can start to address this pain, we first need to bundle all your content into one place where it can be tamed.

There's a lot of chatter around bundling and unbundling lately. These conversations generally focus on unbundling old-line media providers. For example, the unbundling of cable television by broadband internet paired with more targeted streaming services. Or the unbundling of newspapers and magazines by Craigslist and direct-to-consumer newsletter subscriptions.

Bundling and unbundling can also refer to features and functionality. As Benedict Evans observes, there was a time when cars didn't come with windshield wipers, turn signals, or radios. You had to buy the base vehicle and then install those options after market. Similarly, there was a time when spreadsheet software didn't come with printing, graphing, or word counts. Those were separate apps you had to buy individually.

Features that were once optional sometimes become essential. This is the situation we see in reading tech right now. Our users are stitching together two, three, four, sometimes seven different apps across all their devices to take their reading digital. An RSS feed reader. A PDF viewer. A read-it-later app. A web highlighter extension. Readwise. And so on.

Sources: Nicole van der Hoeven & Kyle Stratis

This is why we built the Readwise reading app to support all kinds of content formats, from web articles to PDFs to Twitter threads to non-DRM ebooks and everything in-between. There's nothing revolutionary here: it's just doing the work — a lot of work — to integrate unbundled functionality in a terribly fragmented space.

While this is not exactly the zero-to-one work we typically seek, the cumulative effect is magical. Gathering together all your content that was previously scattered across half a dozen apps on different devices unlocks functionality that would otherwise be nearly impossible. For example, we already mentioned how you can perform full-text search across all your books, articles, notes, and highlights. With everything in one place, you can also begin to conquer content overload by more thoughtfully deciding what to read now, what to read later, and what to read never using various workflows.

You can see the need for this workflow most clearly in the overflowing queues of read-it-later apps such as Instapaper or Pocket. You would think that the more you use one of these apps, the better it would get. Instead, the experience steadily degrades causing would-be power users to churn after a few weeks or months. Why? Because most of us save more things than we can actually get through. As a result, your inbox runneth over, turning your digital garden into a digital graveyard.

We discovered that you can avoid this fate through a game-like workflow that helps you decide what to read now, what to read later, and what to file away, keeping your library neat, tidy, and delightful.

Some of our users have actually backed into this using some combination of folders in Instapaper, tags in Pocket, databases in Notion, or reading lists in Roam. But organizing things this way quickly becomes a chore, like pulling weeds or cleaning your house. So we've embedded a fun, guilt-free triage workflow directly into the Readwise reading app freeing you up to spend more time doing what you actually want to be doing: reading.

During Reading: Getting into the flow

At last, we come to the reading experience itself.

As mentioned above, it's our view that technology has not yet meaningfully improved the practice (or experience) of reading itself. Instead, we're still in the skeuomorphic phase in which words have merely been moved from printed paper to digital screen.[5]

So what does it mean to innovate the reading experience?

Answering that question will require several essays and, candidly, you probably won't be persuaded until you experience the product yourself. But here are some examples of where we're immediately focused.

Highlighting as a first-class feature. It would not be an exaggeration to say that Readwise is built on a foundation of highlights. Accordingly, highlighting is core to our reading app. This is in contrast to every other reading app we’re aware of in which highlighting is always a second-class feature. Neglected. Paywalled. Bolted-on after the fact. With our reading app, you can highlight images, rich-text formatting, and hyperlinks; you can highlight using the keyboard on desktop and quick taps on mobile; and you can highlight any web page using our browser extension.

New forms of annotation. Compared to scribbling a note in a 1-inch margin, software should make it much easier to annotate what you read and highlight. Instead, there are few things more frustrating than trying to type a note on a Kindle or annotate a PDF in Acrobat. Not with Readwise. With our reading app, you'll be able to easily create threaded annotations (like in Twitter or Roam) or wikilink to other documents, authors, notes, & highlights. We believe the best way to read between the lines is to write between the lines. With this in mind, one feature we're particularly excited about is a proverbial "gateway drug" to ease casual users into more frequent annotation. 🙃

New techniques of navigation. One particularly annoying limitation of reading digital compared to reading analog is navigation. If you need to hold your place to flip back a few pages, it's so much easier in meatspace. Place one hand where you are and use the other to find what you're looking for. Similarly, if you need to compare and cross-reference two documents side-by-side, it's so much easier to simply spread out papers on a table. There have been some half-hearted attempts to solve these navigational challenges in existing reading software, but they’re missing a dimension. In our reading app, you'll be able to hold your current place and zip around your documents as if you were playing a video game. Spatial orientation is the key to not getting lost. If you need to compare another document, or simply keep a referenced figure in view, you'll be able to throw up a side-by-side parallel view.

Enriched context & metadata. The Readwise reading app is high performance software for high expectation users. In the same way a modern jet fighter anticipates the needs of its pilot by projecting critical instrumentation in a heads up display, so too does the Readwise reading app enrich the reading experience through context and metadata. This includes parsed metadata from around the internet as well as judicious use of AI-based summarization, auto-categorization, and social chatter to add context.

While these features are exciting, they still dance around the most pressing problem of the entire digital reading experience.

The real reason that most people prefer reading on analog paper books or simple e-ink devices to reading on their phones, tablets, or laptops.

We're talking, of course, about distractions.

With our devices barraging us with a constant stream of DMs, text messages, and app notifications, it's no wonder that most of us still retreat to the safe space of printed paper or e-ink devices when it's time to read for real. Even when we put our phones in do-not-disturb mode, the cue of the device in hand still triggers a powerful chemical addiction urging us to seek the variable rewards of our email inboxes or social media feeds.

As of right now, the best way to get into a deep flow state while reading is to literally unplug.

Unlike any of the features mentioned prior, we'd be lying if we claimed with certainty to have the solution to this problem. What we can say is that we're focused on it and have some promising hypotheses. Like Venkatesh Rao, we remain optimistic that the solution is not to "smash your smart phone and go live in a log cabin to reclaim your attention and your life from being hacked by evil social media platforms." Indeed, we believe that the same technology that undermines our attention can be harnessed for the opposite effect.

If we have one big hairy audacious goal for Readwise, it is this: to create a reading app that helps you enter a deep state flow.

What is this better reading app not?

When describing something new, it’s sometimes helpful to not only speak in terms of what it is, but also what it is not. Accordingly, let us set forth some product objectives we are explicitly NOT trying to achieve with the Readwise reading app — "nongoals", as Joel Spolsky would say.

Not another note-taking app. We are NOT building another note-taking app. There is incredible innovation taking place in this space already and we have little to add but for ensuring your notes and highlights cleanly interface with your tool of choice. (Again, we're using the term “note-taking app” here in the colloquial sense. We acknowledge these apps encompass much more than just note-taking.) Meanwhile, our passion from the start has been reading, and our frustration continues to be the lack of software innovation in this phase of the knowledge acquisition loop.

Not a reading social network. We are NOT attempting to make reading social or enable an internet-scale annotation layer. To be clear, we can definitely appreciate the allure of this dream. As avid readers ourselves, there's nothing more we'd love than to discuss what we're reading with like-minded folks. As entrepreneurs, there's no business that’d be more lucrative than pulling off a new social network. But despite the constant cries for something like this on Twitter, we think it's a classic case of worse is better. You couldn't invent Figma before Sketch and Illustrator nailed the single-player vector-based experience. So it is with reading software. Or so our theory goes.

Maybe we'll be proven wrong. Maybe someone will have an insight that drives a wedge into the network effects surrounding Twitter, Reddit, Slack, Discord, et al. But that group will not be us. As toolmakers, we see so much low hanging fruit in single-player mode that we can hardly fathom multiplayer.

Not a summarization service. We are NOT trying to apply the wisdom of the crowds, artificial intelligence, or mechanical turks to replace everything you read with derivatives. We do believe summaries, auto-categorizations, and social chatter can augment your reading experience. But we don't believe these should be the product itself or serve as a substitute for reading the primary source. We want to help you go deep into what matters. Not skim along the surface.

How are we building our own reading app?

Whenever we read a high-minded thinkpiece describing what software should be like, our natural reaction is one of skepticism: Is this author just another dreamer or an actual doer? We can assure you this reading app is not Project Xanadu. We've already built the proverbial base of the iceberg, and we have dozens of private beta testers using the software for hours a day.

Let us conclude this announcement by sharing a little about our development process and timeline to date.

We've been imagining our own reading app since we started Readwise in 2017, but it didn't make sense to begin building until earlier this year. Importantly, we lacked the right core abstractions. You can't transcend current systems to build a new tool for thought before you've thoroughly probed the idea maze to find the right atomic concepts. In the words of Kevin Kwok:

The best products map to how customers think about their workflow... They choose the right atomic concepts. These are the core concepts around which the entire product is built. They not only align with how customers think of their workflow, but often crystallizes for customers how they ought to. Great atomic concepts are honed and then extended and built upon in more complex compounds that…well for lack of a better word…compound.

In future essays, we'll explain the core abstraction underlying our reading-oriented tool for thought. For now, suffice it to say that we can’t simply cargo cult the block-based architecture of Notion and Roam that’s set off the so-called text renaissance in the note-taking space. Blocks are simply too liquid. This fluidity is wonderful when it comes to writing, but when it comes to reading, it's like trying to eat a bowl of chicken noodle soup with a fork. We needed to come up with our own atomic concept. Something more solid.

We began to grasp our reading app’s core abstraction at the end of 2020, which enabled us to officially begin building in February 2021. Since then, we've developed a local-first web app, cross-platform mobile apps (iOS & Android), various browser extensions, and a series of API integrations. This speed is testament to Martin Fowler's Design Stamina Hypothesis that when you find the right abstractions, you can move quickly without constantly painting yourself into a corner.

We can't conclude a discussion on how we're building this reading app without also mentioning who is involved. As you can probably imagine, the scale of this endeavor vastly exceeds the two of us alone. We’re thrilled to be joined by eight other teammates, all of whom share our passion for reading technology.

We'll briefly mention them here, in chronological order of when they joined Readwise: Meghan Mirchich on finance & ops; Tadek Teleżyński on backend development; Angie Nguyen on customer success; Artem Litchmanov on full-stack development; Adam Lynch on full-stack development; Erin Moore on community; Mati Tucci on front-end development; and Jesse Bennett-Chamberlain on design. Stallions...

Wrapping up

As of right now (September 2021), the Readwise reading app is still in private beta. We’re not limiting access as some invite-only strategy to manufacture a sense of artificial scarcity, perceived luxury, or intentional community. Those can all be effective tactics for generating buzz, but in the same way that social is not in our DNA, exclusivity is not our style. Our reading app is in private beta simply because it's not yet ready for self-serve onboarding. But it's getting there rapidly and we're letting in friendly beta testers as quickly as we can!

If you're an existing Readwise subscriber, you will be among the first users to gain access to the public beta once it's launched. If you wish to gain earlier access (or you’re not already a Readwise user), sign up to join the waitlist for the private beta.

We’ve been working on Readwise since 2017 and with this new reading app, it feels like we’re only just now getting started.

In the immediate term, we could not be more excited to finally be in position to deliver all the features our users have been asking for: from instant syncing, to highlighting images, to powerful search.

Longer term, this reading app is the future of Readwise and we’re cautiously optimistic that it’ll pave the way for a whole new category of software: tools for thought built not for writing, but for reading.

We thank you for joining us on this journey 🙏

  1. We're too impatient to also be futurists. We want to make something that people love now. ↩︎

  2. 2 x 2 x 2 = ~10. Technically, 2 x 2 x 2 = 8 and we need to achieve a 2.15x improvement on each stage for a 10x, but who's counting? ↩︎

  3. Reading retention is a tiny, non-venture scale market. Would not recommend. ↩︎

  4. 50% "Very Disappointed" in case you were wondering ↩︎

  5. There are, of course, some notable exceptions. For example, instantly looking up the dictionary definition of a word in Kindle is a game-changer for many readers compared to reading on paper. ↩︎