Sundog Bio News | August 2025
Good morning,
The Sundog team have spent August getting ready to build version 1 of the Sundog product. You - and our potential users and customers - have seen our prototype. Now, with the summer nearly over, it’s time to start building for real.
To help us do that, we’ve taken our first tranche of external investment. This is incredibly important to us. It means we can focus entirely on making Sundog, and early next year release a product that’s designed for and loved by scientists.
In further exciting news, we’re so happy to welcome Richard Tunnicliffe to the team. Richard has over a decade of experience as Principal Engineer at companies including SwiftKey, Privitar, and Emitwise. He’s solved difficult problems across a huge range of domains - including, in the last month, ours. You can read more about that in Tom’s update.
Alex Mitchell | CEO update
It’s great to have our first round of investment in the bank. It gives us freedom to focus on delivering an excellent product; to spend time talking with our customers and users, solving the difficult technical problems that come with our domain, and designing the tools and interactions that will make work better for scientists.
Our investors are, themselves, people whose own work we have followed and respected for years. Having them on our cap table brings us a wealth of expertise and support, as well as strengthening our network. The support of this incredible group is so important to us as we build v1 of Sundog.
In August, we also received Advance Assurance from HMRC. This means that Sundog qualifies for SEIS (and later EIS), giving our first investors tax relief for their support. Anyone participating in Angel rounds, or our current Friends and Family round, can be confident that their investment will qualify. If this is interesting to you, I’d love to have that conversation - you can reply to this email or email alexandra@sundog.bio.
We’re going into September in a great position. We’ve got investment; we’ve got a brilliant community of investors; and we’ve done some important development work on the foundations of the product. I’ll let Tom tell you about that.
Tom Armitage | CTO update
This month we made our first public code release as a company. It’s not the primary product we’ll sell to customers; instead, it’s an open source release: a plugin for the OpenSeadragon library that powers all our “slippy” image views. It’s available on GitHub right now.
The prototype I’d built in Q2 had confirmed that our approach to images was going to be critical to the Sundog experience. It also revealed the impact our initial approach was going to have, longer term, on development, maintainability, and cloud infrastructure bill. The plugin is part of our answer to the question “could we do this better?” I’d like to dive into that answer in further detail, because it’s a great example of the shape of problems we are working on at Sundog: it reads like a complex technology problem, but at heart is about delivering a great product experience, whilst also making something that scales (technically and financially) and that we can work with into the future.
The images users interact with in Sundog are “slippy” - you can drag, pan, and zoom them fluidly with a mouse, zooming right in to their original size with no loss of fidelity. Our slippy images are - were - stored in the “Deepzoom” (DZI) format, which tiles the image at a number of zoom levels, very similar to how online map tools work. A DZI for a single image is a directory containing potentially thousands of smaller files, along with a small metadata file. The right tiles are streamed into the browser as a user moves around the image, minimising the bandwidth needed to load and explore the image. It works really well, and it’s an established standard.
But: that’s a lot of files for each image. On our back-end, moving or deleting an image now involves moving or deleting thousands of files, which translates into thousands of actions on an object store like S3 (Amazon’s file storage system and standard). And on S3, each action has a (tiny) price. As we start manipulating a lot of images with their respective DZI directories, we start running up a bill. If we had fewer files per image, it’d be cheaper - and faster - to manipulate those files in our object store.
Our SZI Tile Source plugin solves this problem, allowing us to work with a single file for the slippy data on S3, whilst still using the streaming-as-you-need-it approach in the browser. It’s made possible because:
- the SZI format exists - essentially, a DZI file bundled into an uncompressed zip file. The image manipulation libraries we use can already generate this, just as quickly as they can make a DZI folder;
- OpenSeadragon itself has a plugin architecture that can be used to add new tilesources;
- S3-like object storage platforms allow modern browsers to make ”Range Requests” to it, where the browser asks for just a small fragment of a file. We we could just ask S3 for the “ranges” of each of the individual files inside the SZI as they’re needed. The user would be downloading the same amount of data, and it’d feel just as quick.
We’re not the only people to think about this: the OpenSeadragon repository has a thread requesting support for single file images going back to 2016!
I could see the value in us solving the product for ourselves. After an initial spike, I handed the main development of this to Richard, our new principal engineer. It was a great way for him to get a feel for the domain and material of Sundog: technical detail about images, cloud services, and integration with esoteric libraries. He sunk his teeth into it, and delivered a fantastic piece of code - not to mention superb documentation. In the process, he’s become our first employee - and an integral part of our technology team.
I’m really happy that this is in the world. Open Source is hugely important to us: it’s the foundation of the operating systems we deploy to, the browser and server libraries that make our work possible, and the many scientific projects that have provided inspiration along the way. If we’re going to consume that work, it only feels right that we contribute back to it. It’s also a marker for the kind of company I’d like Sundog to be: one that participates, shares, and contributes with the ecosystem around it.
Which is why, before we’ve even shipped a product, we’ve shipped code, shared it with community, and plan to maintain it for some time to come.
We’ve had a summer of exploring - with our potential customers and users, as a team and with the technology we’ll be using to build the first iteration of our product. We’re excited, in September, to start building the foundations of the product itself; to start making real our vision of making work better for scientists.
We’re so glad you want to follow along with us. As always, if you want to talk with us - just reply to this email or get in touch with alexandra@sundog.bio.