Hello,

It’s been an important month for Sundog. We’ve onboarded our first customer and started a new phase of our product development. Today, we talk about what we learned from our launch; from our customers, and from the work we did to get our product to them.

Alex Mitchell | CEO update

Alexis Barr’s Cell Cycle Control group at Imperial investigates how cancer cells have lost control of replication, to identify points of intervention that could prevent tumour growth. They’re Sundog’s first customer.

We went over to the lab to set them up with Sundog in-person. Onboarding a lab to Sundog isn’t just setting up user accounts; we want to set our users up for success. It was important that we could discuss with them the uses Sundog might have for their lab, and tailor our guide to features and functionality to those ends.

The lab wanted to use Sundog in far more ways than I expected. When Alexis agreed to buy Sundog, she told me about a specific use case that we would help with. Her lab collaborates with pathologists who work in clinical settings. They currently do that using PowerPoint and Zoom - makeshift solutions that don’t make the best use of anyone’s time or expertise. I went into our onboarding session expecting most of the questions we got to be about external collaborations on two-dimensional pathology images.

We certainly got those questions! But the lab also talked about use cases beyond what we’d heard about in the sales process. One of the team was flying to Japan for a month, to learn a new imaging technique. They talked about how they could use Sundog to both share the images they acquired with the team in London, and to troubleshoot with the team in Kobe when they acquired images at their home lab.

Some questions made us question the assumptions we’ve made in the last six months. I’ve told plenty of scientists that we don’t yet support time series images, because our interface won’t support the hundreds - sometimes thousands - of images that this work generates. That’s something I repeated in our onboarding session; “we don’t do this today, and it’s on the roadmap for the summer.” In response, I heard something new; “well, we’re normally only interested in seven or so frames, even if we take hundreds of photos. This might not be perfect yet, but it works for us.”

Alongside remote and international collaboration, though, one of the most important ways the Cell Cycle Control lab wants to collaborate is with each other. We heard the clear need for alerting lab members to discussion and questions quickly and easily - which has helped prioritise working on notifications and alerts in April.

Throughout our sales conversations and UX testing, it’s been exciting to watch people think with Sundog: applying our technology to challenges that we didn’t know about before. For the first time, we saw that thinking with from a paying customer, who will use Sundog in ways we didn’t expect, to do things we haven’t thought of. I can’t wait to see where that leads our customers, and our company.

Tom Armitage | CTO update

The Sundog launch was highly unexciting from a technical perspective. No launch-day drama, no surprises; we invited the lab members to the production product, got them signed up, and they started poking the product. And nothing broke.

We watched one lab member upload hundreds of files, and the site Just Worked, our image processors spinning up in parallel to prepare their images for collaboration. I waited for the errors to appear in our logging tools and… none came. As I demoed the StackViewer, six cursors and avatars popped into life and started following me around the image. Just great.

What made our launch so effective?

We’ve worked hard to make sure that there is almost no difference between development and production tools. Our software is deployed against a number of AWS products. Instead of making developers run their own AWS environments for development, or compromising and developing locally against very different tools (always a good way to generate “works on my computer”-type errors), we use Localstack to emulate AWS services locally - and we provision it with the same Terraform scripts we provision the main product with. This helps to minimise the difference between live and development environments, which makes us more confident in our infrastructure. When we were building this up in the autumn, it felt like a lot of work to get off the ground; I certainly second-guessed the level of effort at the time. Was this too much, too soon? It turns out that it wasn’t, at all. That work has paid off time and time again: a good investment.

We also chose not to launch the product too early. We originally targeted the end of January for launch. As that approached, it was clear we could make that deadline if we chose to release some key functionality - support for “Projects” - post-launch. We chose not to do this. Projects provided ways for users to organise their experiments, but more importantly, made it more feasible to control access to experiments in granular ways. This changed how the product worked in some fundamental ways. Launching Sundog only to have to re-educate our users on it a few months later - effectively, having to launch it twice - was not a good use of our time. Better to complete Projects to our satisfaction, and launch when it was ready.

We were confident in the low technical risk of launching in March, but launching to a single lab, with a handful of users, emphasised the reputational risk of not getting our launch right. If the product wasn’t up to scratch, or was unreliable, we could lose those early users, and cloud the reputation of our infant product. We made sure we had time to sweat the small things in the run-up to launch, sanding the rough corners out of the UI and UX of the product, and making sure the documentation was up-to-date.

April is going to be different to the past few months. We deliberately have no major work planned. Instead of grinding away at tickets, we’re going to do two things:

• we’re going to listen to our users, fix obvious bugs, and work out how we can meet the needs that emerge from the two labs using the product.
• in parallel, we’re going to explore, prototype, and think about what could be next, culminating in a planning session at the end of the month.

Playing, researching, and exploring is how product invention happens. We need to think about big swings, new features, and big improvements.

Richard’s exploring improvements to our imaging pipeline and metadata processing. I’m looking at some meaty new interactions in the StackViewer, as well as ways of notifying users about updates. We’re working on a new company website, spending time exploring how we communicate the product, and its progress.

And Alex and I are working on prototyping the process of selling, how we explore customers’ images and present back to them in order to bring them on board. This is particularly exciting now that we are no longer selling a hypothetical product: we’re selling something real, that you can have today. We can set your lab up in an afternoon.

Spring is a time of renewal, and we’re using the time to reset after some busy, busy months. By most standards, we were not crunching, but it was a long period of committed work, and the pressure builds up. April is for releasing that pressure, and working out what working on Sundog with Sundog in the world now looks like. Doing more, smaller, things, more often.

We’ve learned a lot this month, and we look forward to using that to make a better product and a better company. As ever, if you want to talk more about what we’re building and how, just reply to this newsletter or email alexandra@sundog.bio.