Custom CSS

Meet John Mitchell: The Head of Product for Scholarcy

Photo of the author who wrote this blog post
Jessica Rachid
7 min read

John Mitchell is the Head of Product for Scholarcy, an EdTech startup that helps students read and understand academic texts more effectively. I was lucky enough to sit down and discuss John’s experience designing the software that is helping students in Higher Education worldwide.

JMR: Can you tell me a little about yourself and your journey into Scholarcy? Maybe how a ‘workbook project turned into a delightful ADHD-fuelled tangent’, leading to your current role?

JM: I had been burnt out from working in tech and running a business, so I moved to the country and started working off the grid in Australia. That was where I began my self-development journey because I wasn’t happy. I was doing so much research for myself, and a good way for me to learn was to explain my findings to others, so I created a workbook.

I had an initial draft, and each chapter became research-focused. After a year, I returned to the first chapter and thought it was complete rubbish. None of my research was evidence-based. I also wanted to figure out if there was a more efficient way to summarize what I’d written.

I used different summarizer services, which is how I found Scholarcy.

I wanted to see how the output from this software compared to my own summarization of my workbook. I was mostly happy with the result and the user experience, but I  injected my own style and scripts into Scholarcy to make it more usable for me.

One day, I decided to code and redesign a couple of pages in the Scholarcy app, including the main library page, and the flashcards. That was when I thought it would be cool to work with this startup, so here I am, with no workbook but working as the Head of Product!

JMR: As a huge fan of 'Tools for Thought' by Howard Rheingold, how do you see Scholarcy fitting into the broad remit of these tools and ideas? Any strong opinions you’d like to share about the state of academic research and science communication?

JM: The whole idea of ‘Tools for Thought’, is how can we augment our thoughts in different ways?

One of the biggest tools for thought is language and looking at language as a technology.  That is one of the tools that have brought us into the kind of thinking that we have today. It also ties in well with large language models and what capacity they have.

Over the last year, I have been hands-on and learned as much as possible about machine learning, machine language models and AI, identifying what the key differences are. As far as I can tell, almost everything is being marketed as artificial intelligence today. But what it actually means is this huge statistical model that is trained on a mountain of data. It is fine-tuned by humans who prompt the user with the modern version of text prediction.

JMR: Let’s talk more about AI hype. What’s the real story behind Scholarcy’s AI and its impact on education?

JM: One of the key functions of a large language model (LLM) is a compression algorithm. Thinking of it almost like a file. You put a bunch of files into a zip folder, or you get a video that you’ve shot in a raw format, say from your iPhone and then upload it onto YouTube or any social media platform, and it compresses it and the file size is a lot less. Therefore, a lot of what LLM is doing in the context of summarising information is a text compression algorithm.

If you had a paper that was 5,000 words and then you had to summarize the text, using Scholarcy, it would show you a 200-word summary. You may have compressed the text but has the user compromised their understanding?

Over the last two weeks, I have been debating with my coworkers about the use of the summarisation tech. The difference between velocity and brevity and how users come to understand the entire text using the software. I appreciate writers who are concise and articulate. They have an economy of words to convey their argument instead of wasting the reader’s time with a lot of unnecessary information.

With text compression, the real benefit you are getting is speed. You are compressing down the text from 5,000 words to 200 words, but how much understanding do you lose from all of the bits you have cut out from that compression?

You need to be concise when conducting research, but you also need to understand what it is you are reading.

Scholarcy is building that bridge for the user by creating software that promotes both understanding and conciseness. Something we have been discussing recently is the idea of inferential gaps. Are academic papers structured in such a way that they help readers understand the paper better? One of my aims is to move away from simple summarisation and cutting down the text into bitesize pieces, and help readers with each step of the research process. I want to give users the option to have the text explained to them at their current level of understanding, whether they are a general reader, an undergraduate or a researcher.

JMR: Efficiency vs. Efficacy: How does Scholarcy balance the act of knowledge compression with knowledge enhancement? Are there any approaches you’re sceptical about?

JM: One thing we're going to be introducing is an overhaul to the flashcards, which is going to be a huge upgrade to the experience for Scholarcy users. When they open the flashcard, readers will first see a structured summary, broken into sections that conforms to the IMRaD structure of many research papers.

Users will have four main sections: Introduction, Methods, Results, and Discussion. They will be able to have the summary presented in a variety of ways. We are also introducing prompts so users can ask questions of the text, and they will be able to dive into the full text and easily jump between sections.

JMR: How does User-Centered Design feature in what you do at Scholarcy? Could you share some benefits and maybe an anecdote from your design process?

JM: I have strong opinions about design because I am using a lot of different apps. It is uncountable the number of pop-ups that come up when I am using an app. User-centred design is about being in touch with other people, instead of thinking about them as an ambiguous demographic.

These are real people who are using Scholarcy.

I only work on products and build software that I am a user of because I can place myself in the shoes of the customer. I know that I am using the product, and so it has to meet my expectations.

JMR: Scholarcy aims to make academic research more accessible, including to neurodiverse users and those with accessibility needs. How do you ensure inclusivity for these diverse user groups?

JM: This is the first time I have worked on a product built for students. The timing has been interesting because I only found that I had ADHD recently.

I have been learning a lot about neurodiversity. Over the last year, I’ve been doing ADHD coaching and researching the difference between my brain and neurotypical people.

Scholarcy has a user-centred design and as someone who has neurodiversity, I look at the structure, the flow, and the presentation of information. One of the big things that I use Scholarcy for is the presentation of content as flashcards, these blocks of text are presented linearly. I will export my flashcards into a CSV file and then add each section as an Obsidian and Knowledge canvas. Students can export their flashcards into their app of choice, and I think this optimises readability and understanding. There are inconsistencies across different academic journals, documents, articles and web pages in general, therefore, having all this disparate information captured and presented in a consistent format and a predictable format can be very useful.

Something I have been working on for a while now is better readability options. When you’re in a Scholarcy flashcard you’ll be able to select from a drop-down menu, and switch between three different themes: lap mode, a calm reading mode, and a dark mode. Users will be able to toggle between small, medium and large font sizes. They will be able to change the spacing of paragraphs and text and then be able to customise and tailor their reading experience. Another important feature is supporting different kinds of input and output formats, for example, text-to-speech. Something that could be a huge benefit for users is having the flashcards narrated and read out loud by a human-sounding voice.

It has been hugely beneficial working with Oliver Back, the Customer Support and Community Manager, because of his background as a Specialist Autism Mentor and Assistive Technology Trainer. He is on the ground talking directly with our users every day, putting together monthly reports that come out of customer feedback which helps us identify the pain points. Together, we build pitches for product ideas and then we prioritise the main features. At least, we have one person on the inside who has ADHD, which is a good thing,

JMR: Where do you see Scholarcy heading in the next couple of years, given the fast pace of technological advancements and changes in how academic content is consumed?

JM: It is early days for large language models and tech in academia. At Scholarcy, we are focused on augmenting what already exists, putting a different structure and user experience around it. We are probably moving away from the trajectory of AI summarization by trying to make the information as concise but also help students really understand the information that is in front of them.

We are building software by approaching the product step by step and not focusing on one particular audience.

JMR: What advice would you offer to aspiring product developers who are eager to work at the convergence of AI and educational technology? Maybe share a piece of wisdom you wish you had known earlier?

JM: It is a different world from when I first started to design products. Over ten years ago, people were still using Photoshop, and handing a PSD file to developers who translated all of that. Now you have a dedicated designer, a dedicated engineer, a dedicated front-end developer, and a dedicated copywriter. Companies have to be multidisciplinary to succeed in successfully creating a product and serving their audience.

One of the biggest differences over the last decade is that designers used to design, and they never bothered with coding. They never had to translate any of their designs over to a web application or a website. The new Scholarcy website is now built in Webflow because it allows a designer to complete the whole process end-to-end.

John’s belief in continual improvement of the user experience is pivotal to his role as  Head of Product and puts user-centred design at the heart of product development at Scholarcy, His insights into the application of large language models, combined with his dedication to enhancing accessibility for neurodiverse users, highlight the innovative strides Scholarcy is making in the EdTech space. As John continues to lead product development, Scholarcy is going to redefine how academic content is consumed, making learning more efficient and inclusive for all students.