MosaiQ Labs
Integrating AI into workflows
MosaiQ Labs
Junior UX/UI Designer
2023 - 2025
At MosaiQ Labs, I designed UX/UI for an AI-driven research platform in close collaboration with developers. My work focused on creating clear, trustworthy AI interactions, reducing cognitive load, and helping users confidently work with complex systems.
When I joined Mosaiq Labs in 2023, the company intended to create an environment that would facilitate users’ workflows by integrating an AI Assistant to support in the database analysing and in speeding up the production of reliable reports. It was exciting working on this project, as it provided a clear view of companies’ perceptions of AI and the possibilities for integrating it into their bespoke workflows.
We had access to our Beta testing users’ feedback; however, we never had direct access to users, which limited the heuristics we could use to gather valuable insights.
EMPATHYSE
Competitor Analysis
Heuristic Evaluation
Insight Definition
User testing
IDEATION
Brainmapping
"How might we..."
PROTOTYPE
Wireframing
Testing
TEST & VALIDATION
Modify the design
UI
Our target audience was primarely finance consultants —highly skilled at understanding market dynamics and driven by making confident, high-impact investment decisions for their company and clients.
From data collected through feedback and direct observations, we were able to define a pattern of repetitive, mechanical tasks that would benefit from the integration of AI.
-
Collecting & storing a large volume of documents in one place
-
Synthesising large volumes of data
-
And organising information
We translated the insights into "How might we" to reframe users pain-points and friction points into opportunities for innovation.
Blurred image, due to NDA, of the Persona we created as our target audience.
CASE STUDY
-
How might we provide an environment for users to collect, add, and share documents all in one place, and in one click? (if possible).
-
How might we offer a quick way to order, or filter, the collected files?
And digging deeper...
We designed the Project Page as a database, showing the list of articles on the left side of the screen. To quickly skim through the articles, or to order them for a quicker search we introduced:
-
Filters at the top of the page to highlight only the required files
-
Tabs at the top of the list, to order the files according to the selected criteria for faster browsing
-
Visual feedback: when the platform fails to upload a file, it provides visual feedbacks in form of a pop-up window at the bottom right of the screen informing the user of the failed attempt, and a red "x" sign appears next to the file's title
-
To "Add Content" to the project, or to "Share" the project's content, we added two dinstinc CTA at the very top of the page, next to the search bar.
-
However, through the download of the MosaiQ Lab extension, users were also able to add multiple pages to their project externally from the platform, which was kind of handy during the collecting material phase of the research process.
Eventually, the updated version of the Add Content menu included the possibility to add a broader array of files to the project
MosaiQ extension open, with some of the opened tabs selected
“I just know where to put all my data and analyze it, which makes my process much smoother"
"Whether I'm preparing a financial report or gathering passive insights for future reference, MosaiQ ensures everything is in one place. The ease with which I can transition from passive data accumulation to active analysis has transformed our workflows."
Samantha, Financial Associate - Finance
“I’ve accumulated data and been able to produce investment memos 5x faster"
"From accumulating insights on emerging sectors to diving deep into potential investments, MosaiQ is my go-to. It organizes both my data and my thoughts, helping me pivot from passive understanding to active decision-making effortlessly."
Jasmine, Associate - Venture Capital
Once we tackled the issue of collecting documents,we focused on the need to synthesise large volumes of data, hence on the users interactions with the AI Assistant. And a whole new set of questions came up.
How might we offer an easy and direct access to automatic AI insights, for each individual file?
How might we design intuitive and effortlessly ways for users to interact with the AI?
How might we incorporate ways for users to view the source from which the answer has been extracted from?
How might we offer a more natural, organic experience of interacting with the assistant?
Inspired by the well know Gmail, we've designed a UI in which each file can be selected, selected as favourite, viewed in its original source, and analysed by the assistant.
To emphasise which file was selected, we changed the background colour for the selected files to strengthen the visual cue between selected and unselected articles.
Lastly, to underline to which files is the assistant engaging with, we added a “Reply to” cue confirming the current AI interaction.
Influenced by the ChatGPT design, we’ve iterated on various of the "reasoning behind the process" versions, and we agreed that the thinking process should be triggered, as opposed to be displayed by default, as it created visual noise and broke the communication flow.
While we iterated on the location of the "thinking process", we ultimately concluded that, as part of the conversation, it would make sense to stay inside the main chat placeholder.
Just like in a face-to-face conversation with an individual, it is normal for the conversation to go back and forth. New information enrich, confirm, or disprove previously gained knowledge, demanding the reshape of the idea forming in our head.
And so it's research. And so the more sophsticated the assistant's answer became, the higher became the need to enable users to ask questions about a specific segment of the answer.
And for the same principle, it became very important being able to trace the context from where information have been extrapolated, and being able to view the content in its original source.
“The perfect companion throughout my research & report building process”
"Working in market research means juggling vast amounts of data. MosaiQ has been pivotal in helping me centralize this data. Whether I'm passively building an industry understanding or actively preparing a report, the platform ensures I'm always steps ahead."
Carlos, Research Coordinator - Market Research