Michal Wiesel is
.
PlatformDesktop & Native Mobile
Time2025 Q1
Trained on each freelancer's profile, services, preferences, and past conversations, the assistant can respond to inquiries, qualify leads, gather project details, and hand the conversation back to the freelancer when their input is needed.
I worked as part of a team of product designers shaping the end-to-end desktop experience — from onboarding and training to conversations, settings, and a new visual language for AI interactions.
Alongside the desktop work, I independently owned the mobile experience, bringing the Personal Assistant and Fiverr's redesigned messaging experience into the native app.
Before committing to a project, clients ask questions, describe their needs, share files, clarify timelines, and try to understand whether a freelancer is the right fit.
For freelancers, that makes being responsive incredibly valuable - but also incredibly demanding.
The opportunity behind Personal Assistant was to make AI useful at exactly this point: not by replacing the freelancer, but by helping them stay responsive, collect the right information, and identify promising opportunities even when they aren't available.
The challenge was bigger than designing a chatbot.
We had to introduce an AI participant into one of Fiverr's most established and human parts of the product, communication between clients and freelancers while making its role, actions, and boundaries immediately understandable.
8.3%More clients converted
15%more clients converted
*Freelancers with an active Personal Assistant.
A conversation could now move between three states:
That meant rethinking more than the chat UI itself.
We needed to make it clear who was speaking, when the AI was active, what it could do on the freelancer's behalf, what information it had gathered, and when the freelancer needed to step back into the conversation.
The experience had to feel native to Fiverr, while still being distinct enough that AI interactions could never be mistaken for human ones.
The onboarding experience to the freelancer introduces sellers to what the assistant can do, how it learns about their business, and what remains under their control.
Instead of treating setup as a technical AI configuration process, the goal was to make training feel both approachable and an immersive celebratory moment: define how your assistant should communicate, give it the context it needs, and decide when it should step aside.
Rather than placing the experience directly inside the existing white inbox UI, we created a distinct, floating chat environment, applying the new visual language developed for Personal Assistant. This helped separate AI-assisted conversations from regular messaging and made it immediately clear when users were interacting within the Personal Assistant experience.
The two experiences remained connected: every conversation handled by the assistant was still reflected in Fiverr’s inbox, allowing freelancers to move naturally between the two.
Within the new chat, we introduced contextual UI elements for structured information, actions, summaries, and handoff moments - turning the conversation into more than a stream of messages, and into a lightweight working interface.
Bringing it to Fiverr’s native app
The challenge was to integrate the assistant into Fiverr’s existing inbox — alongside regular conversations — while translating the new messaging language into a much tighter mobile environment.
not recreating it
I adapted the new conversational patterns to Fiverr’s mobile design system, creating a clear hierarchy between AI activity, client messages, contextual actions, and handoff states, without making the inbox feel like a separate product.
From conversations to business
More clients converted
higher value per order
Beyond answering messages, the product has grown into a broader layer of Fiverr's marketplace — helping freelancers qualify opportunities, maintain responsiveness, measure assistant-driven performance, and connect with clients across different entry points.
For me, the project was an opportunity to work on a new kind of product problem: designing not simply an AI interface, but the relationship between a user, their AI representative, and another human — and making that relationship understandable across an existing product ecosystem.
Product design:Fiverr Product Design Team
Mobile product design:Michal Wiesel