Hi, RH I'm Rafaella!

The Challenge

Life scientists are drowning in literature they can’t afford to ignore. Every result from the bench raises the same questions: does this fit what’s already published? Who has seen this before? What explains it if it doesn’t fit? Answering them means days of searching, reading full papers, and cross-checking findings by hand.

General-purpose AI tools promised a shortcut but didn’t earn researchers’ trust. They summarize instead of investigate. They still leave the searching to the user, they read abstracts rather than full methods, and they produce confident prose with citations that can’t always be checked. Some of those citations point to retracted papers, or to papers that don’t exist.

For scientists, an answer you can’t verify is worse than no answer. A wrong claim in a thesis, grant, or manuscript carries real professional cost.

So the challenge wasn’t just to build an AI that answers research questions. It was to design one researchers could audit: one that shows its work, shows where the evidence agrees and where it conflicts, and is honest about how much confidence each conclusion deserves.

Design challenge: How might we make AI-generated research answers as checkable as a colleague’s literature review, so scientists can trust them enough to act on them?

The Solution

BioSkepsis is an agentic AI research assistant that works the way a careful scientist does. It investigates instead of summarizing, and every step it takes is visible and checkable.

The design rests on one principle: trust has to be shown, not claimed. Every part of the experience gives researchers a way to verify what the AI tells them.

The Research

Before designing, I wanted to understand how researchers actually work with literature today, where existing AI tools fail them, and what would make them trust an AI answer enough to act on it. To validate my hypothesis, I explored existing AI tools that focus in the academic and scientific areas and interviewed active researchers.

AI Competitor Tools reviewed Elicit Consensus

User Interviews: Uncovering Core Needs

I conducted more than 10 virtual interviews with researchers across academia and industry, from PhD students at UK universities to senior scientists at private research institutes. Using an affinity map, I grouped their feedback into five core insights.

Understanding Stakeholder & User Needs

To design a product that works for both the business and its users, I aligned the BioSkepsis team’s goals with what researchers told us they needed.

Stakeholder goals

• Build trust as the core differentiator. Stand out from general AI tools and competitors through verifiable, evidence-scored answers rather than speed alone.
• Grow institutional adoption. Expand from B2C and individual researchers to B2B research organizations, private laboratories, universities and research partners.
• Increase retention. Give researchers reasons to come back beyond one-off questions.
• Reach researchers within their already set work lifestyle. Integrate with existing tools such as Zotero and Claude rather than competing for attention.

The Ideation

Sketches

After defining the core user flows, I sketched key screens to map how researchers would move from a question or result to a verified answer. Because the outputs are dense with evidence, citations and scores, the priority was a clear hierarchy that keeps information readable without hiding detail. Researchers should be able to grasp the verdict at a glance, then dig into the evidence behind it.

From there, I moved to Figma, refining layout and hierarchy in mid-fidelity wireframes before adding BioSkepsis’s visual identity.

Wireframes

Mid-fidelity wireframes let me test structure and hierarchy before any visual styling. The focus was on three questions: how researchers start, how the evidence is laid out, and how work carries over between sessions.

The Visual Elements That Drive The Solution

Every visual choice in BioSkepsis supports one goal: making dense scientific evidence feel clear, credible and approachable. Together, these choices keep the interface out of the way. The evidence is the hero; the design just makes it legible.

The wordmark combines “Bio”, from biology, and “Skepsis”, the Greek word for thinking, reflecting the product’s core idea: healthy scientific skepticism, powered by AI.

The Outcome

With an average satisfaction rating of 9/10, researchers found BioSkepsis especially valuable for life science literature review and result interpretation.

In feedback from over 30 early adopters, BioSkepsis received an average satisfaction rating of 9/10. Researchers highlighted its value for verifying findings, exploring a field in a structured way, and saving time on literature review. Today it is used by scientists at institutions including the University of Cambridge, King’s College London, IMBB-FORTH and the University of Cyprus, and it is available as a connector in Claude.

The Challenge

RsrvIn began as a reservation tool for venues, restaurants and bars. As the platform evolved, its focus shifted to event ticketing for Cyprus’ nightlife and comedy scenes, while still offering places to eat and drink before and after the show. Each pivot added new features, flows and screens on top of the original product, across web, iPhone and Android.

Built by a team consisting only of developers, the interface had no shared foundation. Colors, typography and components were defined screen by screen, so every new feature meant redesigning the basics and the experience drifted between platforms. Reservation-era screens and new ticketing flows didn’t feel like the same product. The brand also lacked the energy of the nights it was now selling. Going out is about excitement, and RsrvIn needed a bold, recognisable identity, especially in the dark mode where most nightlife browsing happens.

On top of that, every step from sign-up to buying a ticket or reserving a seat had to feel effortless and reliable. In a booking product, a confusing screen or an unfamiliar-looking checkout is enough to lose a sale.

Design challenge: How might we build one design system that unifies RsrvIn’s evolving product under a bold, consistent identity across web, iOS and Android, so discovering and booking a night out feels effortless on any device?

The Solution

RsrvIn’s design system is a single source of truth in Figma. Shared colors, typography, components and layouts make every screen look and behave the same, whether someone is browsing events on desktop or booking a seat on their phone.

The system rests on one principle: consistency should feel effortless, not restrictive. Every part of it helps the team move fast while keeping RsrvIn instantly recognisable.

The Research

The founding team led the market research: who RsrvIn’s customers were, which scenes to focus on, and how the business would grow. My focus was the design side. I needed to understand what the product needed visually, how competitors presented going out, and what a brand for Cyprus’ nightlife should feel like.

Understanding Stakeholder Goals & User Needs

To build a design system that served both the business and its users, I aligned the founding team’s goals with the needs of RsrvIn’s two audiences: the fans who buy tickets, and the organizers and venues who sell them.

Stakeholder goals

• Unify the product after the pivot. Bring reservation-era screens and new ticketing flows together into one coherent experience.
• Ship faster with a small team. Stop redesigning basics for every feature, so the founders can build and launch quickly.
• Stand out in a crowded market. Give RsrvIn a bold, recognisable brand that feels like the scene it serves, not a generic ticketing tool.
• Make the platform’s value visible. RsrvIn’s organizer tools are a key strength, but their value only shows once someone uses them. The design needed to communicate it at first glance.
• Prepare for growth beyond Cyprus. Build a system that scales to new markets, scenes and platforms without starting over.

The Ideation

Sketches

After defining the core user flows, I sketched key screens to map how people would move from browsing to a confirmed booking, on desktop and on mobile. Because event discovery is driven by artwork, dates and venues, the priority was a clear hierarchy that lets each event’s poster stand out without the interface competing for attention. Users should be able to spot something worth going to at a glance, then book it in just a few taps.

From there, I moved to Figma, refining layout and hierarchy in mid-fidelity wireframes before applying the RsrvIn design system: its colors, typography and components.

Wireframes

Mid-fidelity wireframes let me test structure and hierarchy before applying the design system. The focus was on three questions: how people discover what’s on, how an event page leads to booking, and how new users get from download to their first ticket.

The Visual Elements That Drive The Solution

Every visual choice in RsrvIn supports one goal: making going out feel exciting and booking it feel effortless. Together, these choices keep the interface out of the way. The events are the hero; the design just gets people through the door.

The RsrvIn mark turns the “in” of the name into a bold monogram, a simple, confident shape that works as both a logo and a favicon. The brand name selection combines “Rsrv”, short for reserve, and “In”, as in reserving in your ticket or booking, reflecting the product’s core idea: from finding a night out to walking through the door, in just a few taps.

Refining for Maximum Impact

With the design system in place, I designed a new marketing website for RsrvIn. Its job was to explain what RsrvIn offers at first glance and turn visitors into ticket buyers and organizers. The site went through several rounds of iteration, shaped by feedback from the founding team and early users, until every page felt clear, on-brand and ready to sell a night out.

Interactive prototype: scroll and click inside the screen, or open it full screen ↗

The Outcome: Mobile App

On iPhone, the design system came together in a fast, focused experience built for booking on the go. A short three-step onboarding introduces what RsrvIn does, which is Book with Ease, AI-Powered Exploration, and Complete Your Ticketing Reservation, then leads straight into sign-up, account creation and email verification. From there, users land in a home feed of events they can browse and book in a few taps, all in a dark interface designed for late-night browsing.

The Outcome: Web App

On desktop, the redesigned web app brings events, venues and bars together in one consistent Explore experience. Search, filters and a map help people find what’s on, while poster-led cards let each event’s artwork do the selling. Event pages keep a clear “Reserve now” card in view, so the path from discovery to booking is always one step away. Reservation-era screens and new ticketing flows now look and behave like one product.

The Challenge

Life science laboratories run on repetitive, high-precision work. Researchers spend hours on sample preparation and liquid handling, track experiments across notebooks and spreadsheets, and juggle shared instruments with no clear schedule. Every manual step is a chance for error, and every scattered record makes results harder to find, reproduce and share. This phenomenon is known as the irreproducibility crisis.

EFEVRE TECH’s AMGEL laboratory robots can automate this work, but they need a control software that researchers without a robotics or programming background could actually use on a daily basis. VITALE comes from Versatile Integrated Technology Advancing Life-Science Exploration. VITALE is specifically engineered to address these challenges by offering an all-in-one solution for laboratory automation and data management. Whether it’s automating routine tasks, ensuring handling of samples, or providing real-time insights into ongoing experiments, VITALE empowers researchers to focus on what truly matters: groundbreaking discoveries.

Building it meant bringing together very different worlds. Robotics engineers, mechanical engineers, software developers, product designers, UI/UX designers and life-science researchers each spoke a different work-language and had different priorities. Hardware and software had to be developed in step, so a change on one side often meant rework on the other. And because the software controls physical instruments handling real samples, precision and reliability couldn’t be compromised.

As project coordinator, my job was to align these teams, keep hardware and software moving together, and make sure VITALE solved the problems researchers face at the bench.

Coordination challenge: How might we bring robotics engineers, developers and scientists together to build a laboratory automation control software that researchers can trust with their experiments, without needing to be automation experts?

The Solution

VITALE is an all-in-one laboratory automation platform. Researchers design protocols, schedule runs, monitor experiments and record results in a single piece of software that controls robotic laboratory equipment, so they can spend less time pipetting and more time on discovery.

The project rested on one principle: automation should give scientists more control, not less. With just 12 months to deliver, every feature had to earn its place by solving a real problem at the bench.

Real-time visibility across every experiment

The Dashboard brings scheduled protocols, active tasks, user actions and AI-driven alerts into one view. Researchers can see what’s running, what’s next and what needs attention at a glance, which improves coordination across the whole laboratory.

Protocols designed for scientists, not programmers

The Protocol Designer lets researchers build custom workflows visually, setting up labware and fine-tuning parameters like liquid transfer volumes without writing code. Scientists get full control over every step, without needing a programming background or specialist support.

A library of proven protocols

Every protocol created in the Protocol Designer is saved automatically to the Protocol Library, building up a collection of tested workflows the whole laboratory can draw on. Researchers can find, reuse and adapt existing protocols instead of rebuilding them from scratch, which saves time and keeps experiments consistent from one run to the next.

A laboratory that runs around the clock

The Scheduling Calendar lets researchers book protocol runs 24/7, with color coding for each researcher and real-time notifications. Shared instruments stop being a source of conflict, and robots can keep working overnight while the team is away.

Every result, searchable and shareable

The Personal Labbook collects data from every experiment into searchable, always up-to-date records. Researchers can find past results in seconds and share selected entries with their team, using templates that adapt to their research.

Coordinating hardware, software and science on a tight timeline

Delivering a hardware-integrated platform in 12 months meant robotics, software and research teams couldn’t work in sequence; they had to move in parallel. As project coordinator, I broke the year into clear milestones, kept hardware and software deliverables aligned, ran regular cross-team reviews, and gathered researcher feedback on each module. This kept everyone working toward the same priorities and turned a tight timeline into a focused one.

The result: a laboratory automation platform that reduces manual errors, keeps the laboratory running around the clock, and puts every protocol and result in one place, delivered within 12 months by a team that stayed aligned across hardware, software and science.

The Research

Before building VITALE, we needed to understand how researchers run their experiments today, what holds them back, and what they’d expect from automation software. I led a mix of quantitative and qualitative research to hear from as many scientists as possible.

How we learned

• Questionnaires with 50 researchers. We surveyed 50 researchers from universities across Cyprus, Greece and the United Kingdom about their daily laboratory work: which tasks take the most time, where errors happen, how they record results, and what they’d want from an automated system.
• One-to-one meetings with universities and institutes. In online meetings with research groups, we explored their workflows in depth. Where teams already used competitor tools, we asked what worked for them and what didn’t. This gave us a first-hand view of the gaps VITALE could fill.
• Networking events, locally and internationally. We presented the AMGEL automation system and VITALE at events in Cyprus and abroad, with our biggest stage at GITEX Global 2024 in Dubai, one of the world’s largest technology exhibitions. I represented EFEVRE TECH at the Cyprus pavilion, running live demos of VITALE, and took the pitching stage to present how AMGEL and VITALE work together to automate research from protocol design to real-time monitoring. The presentation brought many attendees to our booth for in-depth demos and partnership discussions, and their feedback was immediate: they valued how easy the Protocol Designer made it to build experiments, asked for more flexibility in connecting different laboratory equipment, and highlighted the need for compliance features suited to their local regulations.

What researchers told us about existing tools

To understand where VITALE could stand out, I reviewed three leading laboratory automation tools: Opentrons Protocol Designer, Waters OneLab and Biosero Green Button Go. Each solves an important part of the automation puzzle, from no-code protocol design to traceable run records and round-the-clock scheduling. But none brings the full workflow together. Protocol tools are tied to their own manufacturer’s robots, scheduling suites are built for large, complex facilities, and researchers are left moving between separate systems to design, schedule and document their work. The SWOT below maps these strengths and gaps, and the opportunities they opened for VITALE.

Tools reviewed Opentrons Designer Waters OneLab Biosero Green Button Go

Understanding User Needs

To move beyond survey answers, we held one-to-one online meetings with research groups at universities and institutes. Each session mapped a team’s workflow end to end, from planning an experiment and preparing samples to running protocols, recording results and sharing them with colleagues. We noted where time was lost, where errors crept in, and where researchers relied on workarounds.

Several teams already used competing automation or laboratory management tools, which made these meetings especially valuable. Instead of asking what researchers might want, we could compare what existing tools did well against where they fell short in real use. Patterns that appeared across several groups, rather than one-off requests, became the needs VITALE was designed around.

What we learned

Across meetings, questionnaires and demos, one pattern kept coming up: researchers weren’t short of automation tools, they were short of tools that fit how they actually work. Most wanted to design their own protocols but didn’t code. The most time-consuming tasks were also the most error-prone. Shared instruments caused constant scheduling friction, and results ended up scattered across robot software, notebooks and spreadsheets. Because laboratories mix equipment from different manufacturers, they needed one flexible system rather than another tool tied to a single brand.

The Ideation

I brought together everything we heard, from 50 questionnaire responses, one-to-one meetings with research groups, and feedback from live demos at events like GITEX, and grouped it on an affinity map. Five clear themes emerged, and each one became a building block of VITALE.

How might we give researchers full control over automated experiments, from designing a protocol to recording its results, without needing to write code or switch between tools?

Early decisions

1. One platform, not separate tools. We could have built protocol design, scheduling and record keeping as separate apps. Because researchers were frustrated by data scattered across systems, we chose one integrated platform where a protocol flows from design to schedule to laboratory notebook.
2. Visual design over scripting. Instead of a scripting interface for power users, we prioritised a visual Protocol Designer, since most researchers we surveyed didn’t code. Advanced parameters stay available, but no code is required.
3. Prioritising for a 12-month timeline. With one year to deliver, we ranked features by their impact on researchers’ daily work. The Dashboard, Protocol Designer, Protocol Library and Scheduling Calendar came first, because together they cover the core cycle of building, running and monitoring experiments. The Personal Labbook followed in a later phase.

The Visual Elements That Drive The Solution

Every visual choice in VITALE supports one goal: making complex laboratory automation feel clear, controlled and trustworthy. Together, these choices keep the interface out of the way. The science is the hero; the design just keeps the laboratory running smoothly.

The name VITALE stands for Versatile Integrated Technology Advancing Life-Science Exploration, reflecting the product’s core idea: one integrated platform that gives scientists the freedom to automate their own research.

The Outcome

Delivered in 12 months, VITALE brought laboratory automation within reach of researchers without coding skills, and drew international interest from day one.

Funded by the European Union through Cyprus’ Research and Innovation Foundation, VITALE was delivered within its 12-month timeline as an integrated platform for designing, scheduling, monitoring and recording automated experiments. At GITEX Global 2024 in Dubai, it led to strategic meetings with the Technology Innovation Institute, Saudi German Hospital, Middlesex University Dubai and Dubai Municipality’s Agricultural Division, with interest in bringing VITALE and the AMGEL robots into their laboratories. Researchers singled out the Protocol Designer for making automation accessible, and their feedback on hardware flexibility and regional compliance shaped the next phase of development.