Growing up and becoming Kosmik.
Making hard choices.
The last month at Kosmik brought more changes to the way we envision our product, our work and the company itself.
As you may have read in our previous issues, we’re scaling our infrastructure, getting rid of all the paper cuts in the product, and making Kosmik more standard by adding expected features like SSO and more usual ways of sharing and collaborating.
There is a book I love about the Amiga computer titled “The Future Was Here” — it’s a marvellous way of saying that the Amiga, despite its prowess, failed to embody what the micro-computing industry would become. Despite being the first multimedia machine to hit the market and despite being supported by the best (at the time) home computer maker, the Amiga was never able to own more than 12% of the European market — and most of that came from Amiga sold as game machines.
The future was PCs with windows, DOS, a vast software library, and a foothold in the business world. The lesson is simple: the most standard, less complicated will always win.
We are facing a somewhat similar moment with Kosmik. In the last few weeks, we’ve progressed more than in the last six months in terms of growth, value proposition and understanding of our users. We have doubled down on the visual research workflow we had identified, shipped improvements to make the time to value in the app much shorter, and have improved many aspects of our back end. But where are the table stakes features like SSO, email invites and more?
Where are the table stakes features?
Many of you emailed us in the last few weeks to ask us about what we now refer to as the tablestakes features:
Single sign-on (with Google and then other providers)
Invite people to a universe with their email (our sharing flow is too complicated)
Make syncing more robust and predictable
Why is it taking so long for us to add those features and can we accelerate them? Let me give you a behind the scene look at why we have to take our time and how Kosmik is made.
Building a sanctuary for your thoughts
Kosmik started out as an iPad only app that we wanted to be focused on solo use cases. No cloud, no collaboration, no easy way to share the documents you were building. Just an infinite canvas and ways to organize vastly different types of content.
We did everything we could to make Kosmik as private as possible and as personal as possible. We wanted our users to have a sanctuary for their own data. What was important for us:
Canvas performances
Offline mode
Multimedia support (videos)
As we were entering the market with this version in mid-2022, we discovered that our user base gradually shifted from being dominated by iPad users (70% at the end of the summer) to being mixed (50/50 in October of that year) to being mainly comprised of mac users (80% in February 2023).
At the same time, we were observing a shift in user behavior. Kosmik had started out as a “visual note taking app”. We were riding the hype of the tools for thoughts revolution kickstarted by Notion and Roam Research. Obsidian, Muse, Anytype and countless others were also entering the market with specific value propositions. Our first users were mainly overlapping with those user bases. People were looking for new tools and were excited by the potential of personalized hypertext systems, backlinks and spatial canvases.
We focused on those early users and their needs (backlinks, tags, markdown support). However as our user base shifted to the Mac, we saw another user base take over the product. That new user base was mostly comprised of designers and creatives. They were comparing Kosmik to a “personal Pinterest”, a “visual Dropbox”, or even a “browser for assets curation”. Their needs were different. Our growth rate changed, and with it our strategy to get to market.
Meeting our users where they are and making hard choices
Those users needed to use Kosmik in professional settings: in their agency, for clients presentations, product requirement documents, memos, briefings, etc. They had a lower tolerance for bugs and paper cuts and needed more collaborative features. The “Apple only” nature of Kosmik was also becoming a problem. Their clients were on Macs but also Windows machines and, most of all, in a browser.
In April 2023, we made the very hard decision to focus on our web app codebase and to kill our beloved native apps for the iPad and the Mac.
This was in some way the first “CEO” decision I had to make. We had raised money on a plan, and we were shifting that plan heavily based on data but also assumptions. The shift would represent months on work for nothing and another 6 months to switch completely to the new code base.
(Side note: I do believe that one of the great VC superpowers is to constantly re-evaluate the company as a set of shifting hypothesis and to judge wether or not the current set of assumptions makes sense).
Our team was two engineering teams and product/QA/other acting as a glue between them. What would it mean to have only one code base? Should I fire half the engineering team? How long would it take us to have a web app that would be as mature as our native app? Were there any blockers that I hadn’t anticipated in making this switch? What about performances, user experience, etc? Was I making this choice too hastily based on some user feedback and some hunch?
My cofounder and I discussed it at length and settled on a plan:
By May we would have prototypes of a new front end which would help us evaluate the performance gap.
By June we would have the same for the back end and especially the multiplayer mode.
By July we should aim for “front end” feature parity.
We aimed at August for the start of our B2B pilot program
We targeted September for the big release of Kosmik 2.0. This date would be the start of the deprecation of the apple-only apps, the goal was to migrate all of our users by the end of 2023.
Postponing, and launching Kosmik 2.0
As is often the case with timelines and roadmaps, we were too optimistic. It was almost impossible to add an offline mode to the new codebase without recoding the back end from scratch, the electron app had major stability issues and the windows version of that app was even more buggy.
We had to postpone the launch of the new apps to November, and the migration from the old user base to the new “2.0” apps only began in December and ended in February (we’re still getting messages on various channels about people wanting to migrate from Kosmik 1.8.4 — our latest iPad version to Kosmik 2.0).
In January, after a very exciting but tiring 2023 we regrouped as a team. We had rebuilt our product, hired, started to monetize, raised a seed round and kickstarted a proper marketing effort.
But we weren’t (yet) out of difficulties. Our growth rate tumbled, the migration wasn’t working so well and the new app, while much more powerful, still had stability issues.
So we stopped everything we were doing and just used the app for a week straight. We imported hundreds of files, browsed thousands of different websites, battle tested the UI, multiplayer and syncing engine, etc. We came out of that period with a 90 days plan to make Kosmik “perfect”.
Kosmik 2.4 with numerous performance improvements, shown above is the reworked PDF reader with zero lag.
By the end of April we had solved the most painful issues and people were organically signing up again. We relaunched our B2B effort, started to communicate again and our work paid off! We signed more users in May alone than in the three months before that. The new web browser was the thing that got us over the hump.
As we added new users, we were confronted with new issues (like the fact that we had no back up of our infra in Asia which led to dreadful download times and made it impossible in some cases to create an account).
As you may have seen in the previous weeks, we can correct bugs and problems in a matter of days rather than weeks or months. That’s because of those two very long, very hard and tough periods where we focused on getting the UI right and solidified the back-end.
SSO, service workers and web workers (two important technical changes) are all in various stages of development and they will bring speed, stability and standardisation to the app. Kosmik 2.9 and 2.10 will be packed releases:
Arrows to connect objects
A new sharing flow which will enable email sharing (meaning that you’ll be able to invite people that do not have a Kosmik account yet).
New notifications center, sidebar and share panel.
Then we will add a new publish mode, templates and SSO.
We’re moving faster and we know who we’re building our product for.
I hope this newsletter helped add some color to what we’re building and why we sometimes take our time to make sure we’re doing things right.
Until next week —
Paul 🧑🚀