Why We Started Kaleidoscope

We've often been asked about what inspired us to build Kaleidoscope in the first place – of all the problems we could have worked on, why R&D software tooling? This post summarizes some of the learnings that led to the founding of the company. As always, if anything resonates with you, please reach out!

Summary points

• R&D drives the world's most crucial technological advances (medicines, new materials, etc.) but the software underpinning much of this work (collaboration, decision-tracking, data mapping, etc.) is not built to handle modern-day processes or scale.
• Existing solutions are not sustainable; stitching together a slew of generic software either doesn't work or breaks at scale, while building custom in-house software is extremely expensive and hard to maintain.
• There is increasing pressure for new solutions today, driven by three main trends: (1) heterogeneity of scientific teams, (2) decentralization and rate of data sharing, and (3) the rapid increase in the role of computation and automation.

Backstory

“Oh yea, this is a huge problem. We've spent the last four years and >$100M working on solving it internally”. The more we spoke to people in the industry, the more surprised we were at how common and consistent the answers and complaints were, from small research teams to the world’s largest pharma companies.

We first observed the problem while Bogdan was working on his PhD, which involved highly collaborative work across a consortium of academic labs and big pharma, and then again while he was working at an early stage biotech: it was nearly impossible to get a clear picture of who was doing what in an organization, where data sat, why experiments were or weren’t done, and how all the scientific projects were interconnected. Even answering a straightforward question like “how did we get this figure/result?” was extremely difficult.

The ability to reproduce scientific findings and rapidly build off of each other’s work are foundational to the field. Decisions to run clinical trials, costing hundreds of millions of dollars, critically hinge on the connectivity, traceability, and accuracy of scientific results... but there is no good software platform servicing this today. This led us to wonder how we could use the convergence of our experience designing world-class products (at both 0-1 start-ups and companies like Google) and doing cutting-edge science, to build a product that elevated the scientific work being done by every organization in the world.

We started Kaleidoscope with the vision of building version control for science, to make R&D more collaborative, reproducible, and scalable.

A tool to capture and contextualize all the elements related to scientific work — tying together things like key experiments, decisions, milestones, data streams, and software packages, all in one place. By building this framework, Kaleidoscope enables scientific teams to structure their processes and capture key metadata as well as collaborate more seamlessly across specialities. The value of this is hard to overstate: collecting, harmonizing, and searching across metadata, for example, is at the heart of the $100M+ BigPharma push mentioned above. It's our belief that no company should have to sink massive resources to recreate the wheel, and it's our aim to use our experience creating world-class software to build a product that is loved by millions of scientists and R&D teams across the world.

Cost and failure of existing solutions

Something that compelled us to start Kaleidoscope was the high cost of this disorganization. Not only are the direct costs high — on the low end: repeated experiments totaling tens of thousands each and huge amounts of wasted time; on the high end: months and often tens of millions lost around inflection events like an FDA process. But there are also massive indirect costs: brilliant minds not problem solving or thinking creatively about the actual science, and a capped ability to leverage previous experiments to generate new findings in a compounding way.

Our mission with Kaleidoscope is to help pave the way to a world where 100% of experiments are reproducible and scientists are spending the vast majority of their time on the science.

This brings us to the next question: how are people solving the problem of planning, organizing, and tracking their science, today? In addition to single Excel sheets (yes, you read that right: some organizations we’ve spoken to use a single, manually-populated Excel sheet to track all of their R&D 🤯), we found that companies fall into one of two groups:

Group A (stitchers): These organizations stitch together (1) various consumer tools such as Asana, Jira, Confluence, Notion, Google Docs, along with (2) ELNs, LIMS, etc. The former is very suboptimal for science — they don’t easily integrate with the bio software stack, the component units (e.g. ‘a task’) are fundamentally different in science than they are in generic project management, scientific dependencies are highly complex, historical record of experimental work matters a lot, etc. The latter is expensive and not built with collaboration, scalability, or closed-loop systems in mind (you can read a short blog about experiment capture beyond ELNs here, or why ELNs will not solve these problems here).

Group B (builders): These companies are building in-house tooling to solve the problem. This is expensive to do — it’s multiple engineering salaries over several months or years, solutions are not productized properly, and the tooling is poorly maintained/not robust to changes in team.

Kaleidoscope helps both groups: for the stitchers, we provide a consolidated, biotech-first tool that centers on the workflows and use-cases that are core to the field. For the builders, we offer a significantly more robust, intuitive product at a lower price point.

Why now

Understanding why this needs to exist today is important. We can boil this down to three main observations/trends in the space:

1. Heterogeneity of scientific teams: Organizations are increasingly employing a more diverse range of people; from clinicians to engineers to bench scientists with various backgrounds, each with their own depth of expertise. Kaleidoscope provides the layer that brings structure and clarity to this cross-disciplinary collaboration, while preserving the context each specialty brings to the table.
2. Decentralization and rate of data sharing: Not only has the volume of data generated increased exponentially over the years, but data is changing hands more and more often. Many organizations are outsourcing various components of their projects, meaning that data is frequently shared between individuals, across teams, and with external partners. Kaleidoscope provides the map of where that data sits, who has interacted with it, and why.
3. Rapid increase in the role of computation and automation: Whether it’s upstream prediction, downstream analysis, or a closed-loop system, it’s becoming clear that wet lab (’traditional’ experiments and bench science) and dry lab (computational work) go hand-in-hand. The best teams of the future will rely on a harmony between the two, and maintaining a single source of truth will be critical in connecting and automating pipelines and workflows. Kaleidoscope is the connective tissue across wet lab and dry lab, integrating with the main tools of each arm to provide a holistic view of the work being done.

We are at the start of this wave in the industry, with the above three trends not only converging in time, but also rapidly accelerating across the field. Problems that were easy to ignore before are now going to be major blockers to scientific output if not addressed. Organizations won’t be able to rely on email, Excel sheets, and a slew of consumer-facing project management tools to do the level of work they need to do, at the rate they need to do it at, if they want to be doing their best science. Companies like Twist Bioscience and Ginkgo Bioworks have been propelling their science through data decentralization and synergies between software, hardware, and lab science (a similar piece here, in podcast form), while start-ups like Spring Discovery and BigHat Biosciences are driven by the idea of closed-loop systems between wet lab and computation.

These trends are also reflected in the strategic decisions that larger pharma companies are making. For example, Genentech brought on a world-class computational biologist to head research and early development and mandate the use of great software across the company. In fact, the promise of powerful synergies between wet lab and dry lab (which will require superb software tools to support it) is a big reason behind massive investments going into parent company Roche. Similarly, Merck hired Dean Li, co-founder of the world’s leading computational drug discovery company, to lead its early R&D through the combination of great software and biology. Others are also following suit, spending hundreds of millions of dollars on better R&D and data tooling, creating huge potential for companies that are working in the space.

Our team

Complex problems require diverse viewpoints and creative problem solving. At Kaleidoscope, we’ve brought together an incredible group of people from a range of backgrounds, to build together. From bio-related PhD's to seasoned software engineers and sharp product minds, we focused on assembling a team that can empathize with our end users to deliver an incredible product experience.

Collectively, we are deep believers in building software that empowers people to do the best work that they can. Biotech/R&D is a space where there is both massive opportunity for improvement as well as massive implications for doing so — from new drugs and better therapies to an acceleration of novel scientific insights, the magnitude of which are hard to predict today. This is what we fundamentally believe and why we’re building Kaleidoscope, no matter how small the first domino that falls.

If you want to chat more about anything we wrote, or you’re interested in finding a way to work together, let us know!