2025 – Ongoing
Product Feature
Creating an on-platform design assistant to enhance overall user experience
CLIENT
Podium
SCOPE
Product strategy, Product design
ROLE
Sole Product Designer on this project
Overview
Unsplash: Cash Macanaya
Users consistently reported that certain Podium workflows felt clunky, with long load times that disrupted their momentum. At the same time, the platform's technical capabilities were expanding rapidly, yet much of that lacked the right front-end pattern to be meaningfully surfaced for users. This presented an opportunity to introduce an assistant for contextual guidance and align it with some growing AI capabilities.
How might we design an intuitive, on-platform assistant that reduces friction and actively optimises users' workflows?
Approach to AI
An Assistant;
To aid users, not take all the autonomy away from them.
Pattern needs to be easily distinguishable from usual UI and Podi.
PODI BRAND & TONALITY
What should this assistant feel like?
Careful consideration was taken to ensure that the design assistant feels a certain way. This guides the interaction and patterns that are developed later on.
INITIAL APPROACHES
I explored 3 main look and feel for Podi, ranging from super friendly to more technical.
The consensus was to pursue direction C, in light of how Podium's outcome tended to be more deterministic in nature. We also wanted the Assistant to feel intelligent and not overly friendly.
DESIGN SYSTEM
The Podi Design System
A library of design components are then created based off ShadCN, with some differentiating
podi COMPONENTRY
I don't know what I can say here
PATTERNS
Establishing fixed patterns
I ensured that there are a few key flows that will work for Podi, so that we did not need to reinvent the wheel every time a suitable Podi feature rolls around. Documenting these patterns also allowed for SMEs to know they can utilise Podi for relevant features.
01
Query Pattern
A familiar AI chatbot pattern, query for Podi is easily accessed from the top header.
It is also a way for user to give feedback on generated content.
QUERY PATTERN
IMAGE GENERATION FLOW
02
Optioneering
Having optioneering in Podi allows for a snappy way for user to explore different outcomes by only running specific solvers.
This pattern allows the back-end system to run on a scratch model, allowing for snappier response time.
OPTIONEERING PATTERN
OPTIONEERING: CIRCULATION OPTIONS
OPTIONEERING: FLEX UNIT OPTIONS
03 [WIP]
Outcome-based Assistant
This pattern generates a couple of possible outcomes based on users' input before they even enter the canvas.
OUTCOME-BASED PATTERN
FINAL UI
BASE44 PROTOTYPE
Learnings
Finding a niche for ourselves
While it is tempting to hop on the AI bandwagon and rush to sell all our features as AI, we wanted to ensure what we have is meaningful.