I get overwhelmed by how much stuff I read online. So I outsourced my memory to an AI app.

On a recent late afternoon, I was having trouble remembering. My browsing history for the day suggested I'd read over a dozen news articles, numerous Slack messages, plenty of Twitter threads, and a bunch of notes for my next assignment. Yet, somehow, I couldn't recall much of it. I remembered some vague contours of the content I had consumed but lacked the details.  

That afternoon wasn't particularly special — a few days later, I struggled to recollect the details of a lengthy COVID story I had read during a conversation with a friend. These instances weren't some crises of memory, nor were they due to a head injury. I just had too much rattling around in my brain. No matter what or how much I read online, my mind can't help but forget it shortly after. I don't blame my brain, either. Most people consume an overwhelming volume of text every day — hundreds of thousands of words — so it's no surprise that our memories struggle to retain more than a few scant details. "Humans have worse memories than we think we do, and memory for text, in general, isn't great," Virginia Clinton-Lisell, an associate professor of educational psychology at the University of North Dakota, told me.  

The internet only makes this brain-capacity problem worse. The online-reading experience is full of obstacles that prevent our brains from locking the information we consume into our long-term memories. When you read a book, things like page numbers and the physical ability to hold and turn pages help your brain make a mental map of the information the book presents you with. Websites, however, don't have those kinds of memory triggers. Because of this, multiple studies found that participants who read offline performed better in comprehension, concentration, and recall than participants who read online.  

The added distraction of the web's business — auto-play videos, pop-up ads, and the round-the-clock avalanche of notifications — certainly doesn't help. "Our capacity to process incoming information is naturally limited," Andrew Dillon, an information-and-psychology professor at the University of Texas, told me. He added that if we try to process too many things at once, "we pay a cost in terms of memory and comprehension or time. There's no free lunch." 

Another problem is that people devote far less mental effort to reading something online since we approach all online reading like it's social media — like it just needs skimming. For information to enter our knowledge structures, Dillon told me, we must distribute attention. "This takes time and effort," he said.  

Because we know we can look up a piece of information anytime we want — whether it's a phone number or George Clooney's birthday — we're far less likely to memorize it. Often during the day, I end up Googling articles I read just a few hours ago because I can't recall more than a few key words.  

To address this problem, dozens of read-later and bookmarking apps have cropped up over the years. Apps like Pocket and Instapaper have amassed millions of users by offering ways to organize links and save what they want to read online. But these apps can feel like a chore to keep and do not ultimately help retain the information on those webpages. Because of those drawbacks, I decided to try out a new, little-known service called Heyday. 

Heyday, which bills itself as an AI-memory assistant, promises to fix the two key challenges I've faced with reading-list tools: it demands little to no effort from me and aims to help me remember things better. Instead of simply cataloging where I read something, it promised to help me recall what I've been reading. In the three weeks I spent with the app, I found it was effective at helping me remember things, but it comes with a catch: Using a memory tool like this has the potential to make your biological memory worse over time.  

A new memory 

Founded in 2021, engineers designed Heyday to act as your memory — it quietly processes everything you read in the background and resurfaces information when you might need it. The app works by automatically scanning everything you look at on your browser: web pages, Google documents, notes, Slack conversations, and tweets. Then, it sorts what you've read into categories based on topic or on how much time you spent on something. Once it's added the information to the catalog, it provides dynamic prompts next to search results or within articles themselves to help resurface the information you've already read.

After installing the browser extension, Heyday went to work scanning everything I was reading. Once it gathered enough information, it began to resurface what I had read. When I Googled "Elon Musk" to look into the news about the Twitter CEO banning journalists, the app pulled up a list of related links from my history with their key summaries next to my search results. In this case, Heyday pulled up a Substack newsletter from a journalist, a tweet thread on how Musk lashed out in a Spaces chatroom, and a profile of another social network people were flocking to. This list allowed me to instantly recall what I've already read about the topic and added helpful context to my search, making it a more valuable use of my time.

When reading an article, Heyday would also underline key words that I've read about in the past, and when I hovered over them, the app told me more about that topic based on what I've read. While Heyday's browser tool behaved like a memory assistant, its website felt like a snapshot of my online memory. It categorized all the content I viewed by type: videos, tweets, research reports, and so on. Since its search tool understands natural language, I could use it to pinpoint a query, such as "articles about Elon Musk's ban," without worrying about the correct syntax or key words. Plus, if there were any articles in my history that I particularly liked reading, I could ask Heyday to store it in a separate folder like "favorites" or "recipes," and the next time I come across similar content, it will automatically recommend I save the article to that folder.  

One potential drawback is that because of how it runs, Heyday collects a treasure trove of data on you and your browsing history. But Samiur Rahman, a cofounder and the CEO of the startup, told me the data is encrypted and that their business model doesn't depend on selling user data or ads. Instead, their revenue comes from the $19 monthly subscription to use the tool. While the subscription model may help keep my data safe, the steep price point limits the tool to people like researchers or journalists who would use it a lot and could justify the cost.  

Another limitation with the tool was that the search widget often took a few seconds to show up next to my search results. So, there were times when I ended up clicking one of the results instead of waiting to see what the pop-up widget resurfaced. And though it never failed to register what I read, Heyday's algorithm occasionally did not understand a link's content and context, resulting in relevant articles not showing up when I expected them to while conducting another type of research. But in my three weeks with Heyday, I consistently felt that it made a difference in how much I could recall — and the experts I spoke with had an explanation.  

New information enters a first "forgetting curve" in our memories, and much of it slips down the drain unless we spend time reviewing the material. Dillon told me that repetition or rereading is crucial to remembering new information better. He told me that Heyday's continuous exposure to the same information can prove vital to aid memory. Similarly, Clinton-Lisell told me that making connections between subjects or themes naturally improves comprehension and memory, adding that if a tool like Heyday helps you make connections as you read, it should enable you to better remember.  

In my experience, because the app repeatedly exposed me to the content I had read and helped me make connections between things I was reading, it helped the information stick in my head.  

Refocusing the mind 

Rahman told me that Heyday's ultimate goal as a memory assistant is to "increase the creative output of individuals." By freeing up the brain from the task of encoding and resurfacing memories, he believes the brain will have more freedom to "focus on things that the human brain is uniquely great at — thinking, creativity, and analysis." 

Heyday isn't alone in this venture. Broader efforts to supplement our memories are underway across the industry. Dennis Xu, a cofounder of the OpenAI-backed Mem AI, a self-organizing workspace, wants to aid the brain in recalling disparate pieces of information so it works less to recall raw data you can easily look up. The goal of Mem AI is to allow people to focus on creative outputs and remember personal memories like a loved one's face. "I think that's a more worthy use of the brain's memory than, for example, remembering what date the Battle of the Alamo took place," Xu said.  

Rewind, a startup that raised $10 million in a round that Andreessen Horowitz led last year, said its "long-term vision is giving humans perfect memory." Rewind captures everything you look at on your computer and constructs a timeline of your activities, letting you simply search for facts and conversations instead of having to remember them or spend time digging through various apps. Personal AI, on the other hand, wants to clone your mind, creating a virtual "second brain" that houses all your memories and data.  

However, if we routinely begin to outsource the job of memorization to internet tools, would our biological memory evolve to become obsolete? There is already evidence that the internet has caused our memories to get worse since we don't feel the need to memorize what we can Google anymore. Personal AI's spokesperson, Jonathan Bikoff, doesn't necessarily see that as a bad thing. He expects technology-enabled minds to be more powerful and reliable than our biological minds. "With assistance from AI, humans may be able to enjoy more of life, embrace forgetfulness, and learn to weave AI into our every day," Bikoff said.  

Though the jury is still out on our memory's biological evolution, Dillon said the extent that the web and digital access is making memorization obsolete, at least for some people, is interesting. "Why learn a poem by heart if you can pull it up on demand? What's the point of learning your math tables if you can just ask Alexa for the answer? Like our bodies, do our minds also need to be exercised to maintain full functioning?" Dillon said, concluding: "I think there's some truth in this."  

I can't help but agree. During my Heyday use, the number of stories from its widget's list I had already forgotten about constantly surprised me. It was a reminder of just how much information I read slips from my brain. But while Heyday was effective at bridging the gap of my limited memory, making research easier, I worry that a reliance on the tool would make my memory even worse. But given the mounting volume of text we read online, perhaps we have already passed the point of no return. The modern world demands that we consume a massive amount of information, and our biological memories simply don't have the ability to remember it all. So instead of fighting a losing battle, an extended hard-drive-esque space like Heyday can be a vital supplement. For me, at least, Heyday is here to stay.

Shubham Agarwal is a freelance technology journalist from Ahmedabad, India whose work has appeared in Wired, The Verge, Fast Company, and more.