Owkin K

Owkin K as an AI copilot for biological dataset queries

Owkin is a Paris-based AI company building tools for biomedical research and drug development. K, now commercially available as K Pro, is an AI copilot that lets researchers and clinicians query proprietary and publicly available biological datasets through natural language.

Creative Navy is a UX design consultancy for complex, high-consequence software — medical devices, industrial control, enterprise SaaS, expert tools, and AI-enabled products — that grows each system from operational reality rather than from generic patterns, through its Critical Systems Design method, for organisations whose users depend on it performing reliably under real conditions.

K is described as a specialised reasoning system rather than a generic AI assistant applied to biology. The underlying model is trained on biology, not the general internet, and operates across multiple data modalities including clinical records, imaging, and genomics.

The commercial question during Creative Navy's engagement was whether K's backend capability could be made accessible to clinicians with low to medium scientific background. K had originally been built around expert biologists, so the interface needed to carry more orientation for users without the same domain intuition.

Discoverability failure caused by invisible capability and bounded data

Users arriving on K did not understand what the platform could do or how to begin using it. The capability was concentrated in the backend: the breadth of datasets, the specificity of biological reasoning, and the ability to answer complex research questions that previously required a data science team to formulate and execute.

Creative Navy's design work treated the problem as capability democratisation. The same system needed to support expert biologists and clinicians with lower scientific background, without assuming that clinicians already knew which biological questions were available to ask or how to formulate them.

K is intentionally data-bounded. It does not draw from the entire internet; it uses specific proprietary, public, and user-uploaded datasets. This made data discoverability part of the core usability problem. Users needed to know what data was available before they could form useful queries against it.

Creative Navy's 8-month Implementation Partnership with Owkin and Merge

Creative Navy joined an Owkin product already in active development. The engagement was not a greenfield design project. Owkin needed targeted UX interventions on discoverability, and those interventions were needed quickly because a product release and an investment pitch were approaching.

Owkin retained Merge as its permanent digital design agency alongside Creative Navy. Creative Navy's role was to establish a design direction that worked for users, while Merge could subsequently implement and extend that direction.

The engagement ran for 8 months as an Implementation Partnership. Creative Navy's involvement included design handover to Owkin's development team, testing support, and QA, extending the work beyond design exploration into the delivery cycle.

Sandbox Experiments benchmarked more than 20 adjacent AI products

Creative Navy's Critical Systems Design method used Sandbox Experiments to test discoverability directions against both user needs and emerging AI product patterns. Creative Navy-recorded engagement evidence states that more than 20 competing and adjacent AI products were benchmarked, including Julius AI and Mindtrip.

The benchmark research identified industry-standard discoverability patterns and competitor features that could be adapted to K's biological research use case. The competitive landscape was used to understand what users already expected from AI research tools and where K needed to communicate its specific differentiation.

Creative Navy explored four topic areas: the Explore page, prompt suggestions, the AI chat box, and dataset presentation. Each topic received 5 iterations. The Explore page moved through a use-case orientation, a prompt catalogue, and then a dataset-first presentation alongside the tool's main modes.

The dataset-first direction became the basis for convergence. During exploration, Creative Navy found that for K's users, understanding what data was available to query was more generative than understanding the tool's abstract feature set.

Data discoverability was prioritised over second-stage prompting

Creative Navy's design exploration deferred second-stage prompting and prioritised data discoverability. The rationale was that users first needed to understand what they could ask before prompt guidance could help them ask it better.

Owkin and Merge participated in brainstorming sessions during Sandbox Experiments. In one session, an Owkin internal expert provided a list of tool capabilities with associated visualisation types. That input grounded the Explore page content in what K actually produced, rather than in assumptions about what a general biological AI assistant might produce.

This decision shaped the later design direction. K's interface needed to make available datasets and main modes visible at entry, because the bounded nature of the data was part of the operating model users had to understand.

Concept Convergence resolved the new-user and power-user tension

Creative Navy's Critical Systems Design method used Concept Convergence to address a central tension in the K interface: new users needed guidance and scaffolding, while power users needed access to complex functionality without friction.

The tension could not be resolved by compromising between the two user groups. Solutions that served new users risked over-constraining power users, while solutions that exposed full complexity risked leaving new users lost. The mode-and-dataset framing addressed the tension by giving new users a structured entry point without removing access to the underlying power.

The iteration count required to settle a discoverability solution took longer than anticipated. Creative Navy-recorded engagement evidence indicates that finding the boundary between sufficient guidance and preserved complex functionality required more exploration than the initial scoping assumed.

Creative Navy's early concepts also created friction with Merge because they operated in a different design paradigm from the patterns Merge had developed with Owkin. The friction resolved once the new direction showed how deeper structural problems were being handled, and the working relationship continued after the project.

Development constraints shaped which concepts entered the release window

Creative Navy and Owkin held two show-and-tell calls per week during the convergence process. Each presentation covered the pros and cons of the explored directions rather than only the recommended option.

Development complexity and implementation timeline were factored into convergence decisions. Several directions that were conceptually strong were treated as post-MVP tasks rather than discarded, because they required more time to specify than the release window allowed.

This preserved the option space while keeping the immediate release scope realistic. The documented case evidence does not describe those post-MVP concepts as delivered within the engagement.

Iterative System Building integrated new screens into Owkin's design system

Creative Navy's Critical Systems Design method used Iterative System Building within Owkin's existing design system. The design system was already comprehensive, and the gaps described in the case were minor missing button states such as hover and pressed states that Owkin was already aware of and planned to address.

Working inside the existing design system matched the timeline constraints. New components and screens were integrated directly into Owkin's design system rather than requiring a separate system rebuild.

Interactive prototypes were produced to show micro-interactions. Mode activation in the prompt chat box required particular attention because the active mode had to be obvious for a first-time user while still feeling like a seamless option rather than an interruption to the interaction flow.

Organizational Integration embedded design reasoning in stakeholder sessions

Creative Navy's Critical Systems Design method used Organizational Integration through repeated stakeholder presentations that included an educational layer. Each design presentation explained users, user behaviour, and how that evidence informed the concepts being explored.

This was not a separate knowledge-transfer session. The educational content was part of every show-and-tell, which helped Owkin stakeholders understand the reasoning behind the design directions and provide more useful feedback.

The twice-weekly cadence also allowed misalignment to be detected early. One session confirmed the direction when Owkin reported that complaints from an internal product launch matched the issues Creative Navy had identified and proposed solutions for.

Client-reported investment evidence and scoped discoverability improvement

Owkin used Creative Navy's designs as the central artefact in an investment pitch. The pitch question was whether Owkin had a user paradigm for making the backend capability accessible, and the prototype was evaluated directly as the lead artefact.

Owkin attributed approximately £5M in investment to the quality of the design work. This figure is client-reported, approximate, and not independently verified; the causal attribution was made by Owkin.

The discoverability problem was significantly improved but not fully solved. Available case evidence describes the improvement as directional, based on internal product launch feedback and Owkin's characterisation of the outcome. No measured user metrics are available.

The strongest documented effect was at the beginning of the user journey, when a new user first encountered K and decided whether to engage. The engagement was scoped to the Explore page, prompt suggestions, AI chat box, and dataset presentation, so the case evidence does not claim system-wide discoverability resolution.

Evidence boundaries for the Owkin K case

The investment evidence in the Owkin K case is client-reported. The approximate £5M figure and causal attribution to design quality come from Owkin, not from independent financial verification.

The discoverability evidence is directional. It is based on internal product launch feedback and client characterisation, and no measured user metrics are available in the documented case evidence.

Creative Navy-recorded engagement evidence supports the 8-month Implementation Partnership, the 20+ competitor benchmark count, the four topic areas, the 5 iterations per topic, and the design workstreams covered. These are engagement facts, not measured user outcomes.