Work
About
Resume
o.wolman@gmail.com

timbr.ai

The timbr.ai platform is changing the way large companies make use of, and benefit from their data by implementing the Semantic Web Standards in SQL. Making it possible for enterprises to conveniently modernize and enhance existing data technologies with knowledge graph capabilities.

The timbr.ai platform is changing the way large companies make use of, and benefit from their data by implementing the Semantic Web
Standards in SQL.

Overview

timbr.ai is the developer of the SQL Knowledge Graph. By transforming databases into knowledge graphs, timbr.ai helps enterprises to develop machine learning algorithms to answer the most complex questions directly from BI solutions.

Project Goal

Redesigning timbr.ai platform from scratch. collaborating with Product & Dev team to optimize the UX in this complex system.

Biggest Challenge

Considering technical & functional aspects while keeping a user centric approach in many super complicated UX challenges.

what i did

UX Research
UX Strategy
Wireframing
UI Design

Role

As a Product designer for timbr.ai. I was in charge of all design aspects, from early stages of UX research & UX solutions to a full UI design.

The Problem

The timbr.ai platform was built by talented programmers with no extensive design knowledge. The team was quick to recognize the disadvantages of existing visibility, such as inconsistency and highly technical design.  I joined the team during a time when they recognized they needed great UX and knew it was time to bring in a Product Designer.

Research & Discovery

Users and Stakeholders Interviews

I began to evaluate the exact design aspects that timbr.ai would need to improve. Because the company is very young and does not have many users, there was no room for quantitative research. Therefore, I conducted interviews with stakeholders and key users.

In the interviews, I tried to understand:

  • Main pain points and frustrations

  • The positive aspects of the product which they enjoy and will want to preserve.

Here is some key evidence I gathered during the interviews:

I think that there is a large range of user level in the system. From managers, data people (scientists, analysts), and fluent SQL speakers. The system is currently very technical and not sufficiently considerate of less advanced users.

Not sure why but there's a tendency to concentrate large flows in one place, wizards with several stages in one screen, separated features are connected or a feature whose parts are scattered in several places in the platform.

There is a lack of explanations, tips, things that will help the user understand parts he might get stuck in. There is a sense that I (the user) am assumed to know what to do. I can defiantly use some guidance.

I think that there is a large range of user level in the system. From managers, data people (scientists, analysts), and fluent SQL speakers. The system is currently very technical and not sufficiently considerate of less advanced users.

Not sure why but there's a tendency to concentrate large flows in one place, wizards with several stages in one screen, separated features are connected or a feature whose parts are scattered in several places in the platform.

There is a lack of explanations, tips, things that will help the user understand parts he might get stuck in. There is a sense that I (the user) am assumed to know what to do. I can defiantly use some guidance.

Results Summary

Problems

  • Far too technical, bad for low oriented users.

  • Screens are too cluttered.

  • Designed with over-reliance on user understanding.

  • Features in the system are not sufficiently incorporated.

To Preserve

  • The tech is amazing

  • Functionally: it does the job.

Insights &
Guidelines

Research helped me discover problems in the previous design, which allowed me to draw early conclusions. From that point, I have defined the guidelines for this project:

Consistency is a key

The system was build in patches. Many flows were disconnected, and there were lots of inconsistencies. Since timbr.ai is an incredibly complex system, consistently is key for creating an easier to use system, and an optimized experience.

Lower the barrier of entry

The platform was designed for advanced users with an over-reliance on user understanding. The system had to take into consideration low-oriented users that are not as technical and needed more guidance.

Divide (or merge) and conquer

The lack of white space was only a symptom for the entire virus. A better UX demanded a rethink and redesign on every feature by dividing or merging different components.

Mind relaxation

There was far too much cognitive overload all over the system’s interface. Many elements lacked spacing. White space is crucial to help the user focus on relevant targets.

The New
Data Modeler

Feature #1

The data modeler allowed users to build a hierarchical model of their own data. The following video is a tour guide for the data modeler.

🔊 Turn up the volume to listen to my narration

UX Decision

Data Modeler

The first thing to approach regarding the data modeler was the structure. I worked closely with the VP Product and the R&D Team Leader on this feature. We realized that graph structure is better for small company but timbr.ai is specialized for big enterprises. Due to that, we created a scalable & super quick modeler that allows optimal orientation on knowledge graphs.

RUNNER up - Graph modeler

  • Bad for large enterprises

  • Not scalable

  • Dictates slow modeling

Winner - table modeler

  • Great for large enterprises

  • Scalable Solution

  • Allows fast modeling

From a Graph
to a Tree

Data Modeler

The following "Old & New" demonstrates the changes we made in visualizing concepts (representation of Db tables).

Old

Difficulty to navigate across large ontologies

New

Fits all sizes

Adding
New Properties

Data Modeler

While the old modeler forced users to go through a lightbox screen for adding properties (Representation of Db columns), the new modeler allowed users to add properties in a much more efficient way. An "adding row" is now always available in the first line and there is no need to jump between screens.

The New
Data Mapper

Feature #2

The data mapper allows users to map a database into the knowledge graph they modeled (in the data modeler). The following video is a tour guide for the map tool:

🔊 Turn up the volume to listen to my narration

Before & After

Data Mapper

The old data mapper suffered from cognitive overload. This led us to decide to combine the 3 steps wizard in a much more elegant way with two imperceptible steps. It benefited non - progressive users who seek more guidance. In addition, it made the overall flow with more breathable layout and more white space.

Before After
before
after

Metadata

Data Mapper

After we cracked the solution for the maptool, it was time to address deeper user scenarios. The selected table metadata allows users learn more about the table, a moment before they map it into a concept.

Main categories. A part of the information architecture

Learn about region's earning trends

Bird's has important financial engine that let independent operators to make money by completing tasks like: charging and moving scooters.

Discovering
Columns

Data Mapper

Another important feature in the data mapper is the ability to do table profiling. Users can dive in and explore statistics and other interesting data behind the table's columns.

Each section specifies a column breakdown & statistics.

The New
Permissions
manager

Feature #3

A complex system, such as timbr.ai needs a tailored solution for managing permissions. The following video is a tour guide for the Permissions manager.

🔊 Turn up the volume to listen to my narration

Restart

Permissions Manager

The old permissions manager was challenging to grasp. It limited users to set roles by defining every endpoint in the system (endpoint ex: "user can reset password"). With my recommendation, we made a pivotal change in allowing users to set permissions by pages (page Ex: login page) and not by endpoints. Also, we used this opportunity to rearrange the system information architecture. That way, users can easily navigate through categories and pages.

Before After
before
after

Creating
New Role

Permissions Manager

Settings permissions by pages goes like this: The user selects a main category in the system, and then chooses if a role can edit/view/none regarding the page. This change really lowers the barrier of entry for a wider range of users.

Creating
New User

Permissions Manager

Creating a new user was never easier. with the new change, all you need to do to add new users is to choose existing roles (default roles or custom). If there's a need for a new role, no problem, create a new role and then assign it to the new user.

Impact & Takeaways

Impact

  • My product design helped the company in different POC stages which lead to better customer acquisition.

  • Many of the UX solutions I provided are getting awesome feedback from clients.

Takeaways

  • I got to take part in end-to-end product design, from the concept stage to the development stage including adjustments according to customer requirements.

  • Worked with top experts in their field, collaborating with product & dev teams from whom I learned a lot about the world of content and work processes in a product company.

  • I've learned to insist where I should and "sell" my UX solutions when I see they fit.