Irene is UK-based product designer.
Email ryna.rudenko@gmail.comLyra. Product design. Early stage.
Lyra
The art folks have always been cautious about technology for good reason, and designing something that felt like it belonged there meant starting from how these people actually work, not from what most platforms expect of them.
The result is a headless platform that combines past interactions with publicly available collector data to help them move faster on the deals that matter, with AI features that handle the overhead so there's more room for finding talent, building relationships, and the work that actually requires their sensitivity.
Adaptable Inbox
One of the suggested features was a proactive relational signal — when enabled, it researches publicly available information about a collector and their background, combined with past interactions, to make more informed judgements on both the legitimacy of a request and the likelihood of a deal.
Columns populated with a sourced information, auto-asking questions to gather additional context, needed to make decisions, without leaving the workflow.
Surfaced shortcuts adapt to each view, helping managers quickly switch between views and actions. Instead of forcing each curatorial site to work around the tool's limitations, we aim to be adaptable to their unique structure and processes, including custom views and generated columns.
There are multiple roles within a single organisation — such as accounts teams and consultants — though in smaller organisations one person can occupy both. This made it important to clearly divide their working views.
Lyra. Product design. Early stage.
Workflows
We created the most common workflow templates to cut hours spent on repetitive tasks every week.
Both "stable" and "pre-gen" blocks on open-ended sourcing tasks are built directly into each specific workflow. Pre-gen blocks run depending on token balance, and I tried to make it transparent and fail-safe.
Lyra. Product design. Early stage.
Experimental / Rooms
I spent a lot of time working with curators, and we often thought about unconventional approaches to art discovery. Our inboxes are busy, promotional emails are rarely opened, and rarely should be. So what if we created a virtual exhibition tailored to a specific audience, using signals from their preferences and past activity? This feature is yet a conceptual one, using ElevenLabs to create voiceovers and cross-references; however, one of the surprising side effects was how it made art ordinarily accessible to disabled people.
Jua AI.
Jua
I concepted the new design system for Jua, exploring how to make their AI-native, data-heavy weather forecasts more legible and actionable. Jua is building world models and has an ambitious goal to create the best physical simulation, and one of the initial steps to do it is modelling the atmosphere. I focused on translating complex probabilistic outputs into clear components, layouts, and interaction patterns tailored for energy suppliers.
Don't be a stranger.
In daily work I appreciate a good amount of restraint, intentionality and scrappiness. Would love to know people, who work with complex or ambiguous areas (like weather/cell modelling), explainable AI.
If you would like to work together, feel free to reach out.