Toric AI

Simplifying Data Pipeline ETL workflows with AI for the Construction Industry

Prototypes used for the Toric AI launch event

Created and prototyped by Renato Ferro for internal and marketing purposes

What

Toric AI is an AI-powered platform that helps construction companies manage their data more efficiently. It automates data operations and makes complex data systems accessible to non-experts, expanding beyond its origins as an ETL tool to serve multiple industries.

I helped transform Toric from an ETL platform to an AI-powered solution, leading product design, AI strategy and development while working closely with leadership to announce the change publicly in just 2 months.

Role

Lead Product Designer
Strategic Contributor
Researcher

Responsibilities

Research
User Experience Design
Visual & Interaction Design
UX Architecture
System Design
Prototyping & Iteration
Team Enablement
Design System

Team

Lead Product Designer
CEO (Acting PM)
CMO
CTO
CS Director
Senior Sales Director

Challenges

This project began before ChatGPT-3 gained widespread popularity.

Timeline: 2 months total
Undefined Vision: Initially, the lack of a clear direction integrating AI features, benefits, and costs.
Emerging Technology: Integrating LLM/RAG and other new AI technologies on an early stage posed significant UX and Dev challenges.
Limited Research and Resources: Limited time and initial customer interviews required reliance on alternative research methodologies.

Outcomes

Developed and launched three AI features, with the first being Entity Extraction for Data Quality:

Data Quality. Set of AI features that help Data Engineers to quickly normalize, classify, and extract data (entity extraction) by assessing quality and detecting errors, empowering even non-experts to boost data integrity.
Data Analysis. AI Features that automates ETL transformations driven by the users desired outcomes, enabling easy and fast insights for customers and freeing up data engineers for complex tasks.
Data Insights. Conversational experience that let users instantly query complex construction project data using natural language, eliminating the need for IT or data analysts.

Overview
Discover
Design
Develop

Problem Space

2. Discover & Define

Compilations of the Discover and Definition Phases

Concept Maps, Workflows, Use Cases, and Archicture.

My Contributions

Workshop & Rapid Approach: Used a fast-paced JTBD research process through call transcript analysis and internal ideation workshops.
Pain Point Analysis: Created journey maps and personas to identify key areas where AI could provide the most value to customers.
Leverage SME Knowledge: Synthesized and analyzed insights from solution architects and data analysts from CS team to shape our AI strategy.

Challenges

1 week: Working within an extremely compressed research timeframe to define and communicate to the CEO potential problems and their solutions.

Artifacts

All the artifacts and diagrams were designed and produced by me.

Problem & Solution Space

3. Design

0:00 Data Quality (Accuracy, Consistency, Health) | 1:20 Analysis (Generative Metrics and Reports) |2:30 Insights (Copilot and Mobile Data Exploration)

Compilation of all final prorotypes presented during the launch event.

My Contributions

I have designed and produced 8 prototypes that showcase the AI capabilities we will be introducing in the product in 2023 and 2024.

Prototypes:
- Data Quality: Accuracy, Consistency, and Health Monitoring.
- Data Analysis: Metrics Generation, Report Generation, Visualization Summarization.
- Data Insights: Copilot and Insights Exploration.
Design System:
- AI Patterns: Created a design system for the overall new AI features patterns.
Research:
- Sprints: Conducted over three weeks of design iteration cycles with the CEO, CTO, and CMO, who also served as the project PM.
- SME Workshops: Leveraged continuous feedback from internal subject matter experts to clarify the business value and optimize user interactions.

Personas

Data Quality

Data Engineer:
Designs and implements database and storage systems.

Data Architect:
Creates data pipelines that transform raw data into analysis-ready formats.

Non-technical users:
Data analysts and Excel users who need to transform data but aren't familiar with advanced tools like Python Pandas.

Data Analysis

Data Analysts:
Extract insights from data and communicate them to stakeholders.

Analytics Managers:
Oversee the analytics function in your organization.

Non-data savvies:
Project Managers, Engineers, Superintends GC, Foremans - Specialty Subs, Safety Managers, Owners reps, and Owners.

Data Insights

Data Analysts:
Extract insights from data and communicate them to stakeholders.

Analytics Managers:
Oversee the analytics function in your organization.

Data Engineer:
Design databases and storage systems.

Data Architect:
Build data pipelines that transform raw data into a usable format for analysis.

Non-data savvies:
Project Managers, Engineers, Superintends GC, Foremans - Specialty Subs, Safety Managers, Owners reps, and Owners.

Challenges

Limited timeframe to align AI strategy with the board of directors and leadership team
Account for technical constraints of Toric's legacy node-based transformation system

Artifacts

All the artifacts and diagrams were designed and produced by me.

Final Prototypes: Quick Walkthrough

Accuracy Prototype

Data Quality

Consistency Prototype

Data Quality

Health Monitoring Prototype

Data Quality

Summarization Prototype

Data Analysis

Autogen Metrics Prototype

Data Analysis

Autogen Reports Prototype

Data Analysis

Copilot Prototype

Data Insights

On-site Data Exploration Prototype

Data Insights

Solution Space

4. Develop

Compilations of Implemented Designs

Use Cases Recorded from the Production Environment

My Contributions

Provided detailed documentation for design handoff to developers.
Acted as a bridge between design and development teams, ensuring a smooth transition.
Adapted Toric's frontend and backend architecture to fit toricai's requirements.
Delivered a design system with developer-friendly components to accelerate implementation.

Challenges

After the implementation of the entity extraction MVP, we began transitioning from Toric to Datagrid.

Ensuring architecture alignment while introducing new features.
Adapting complex AI features into a user-friendly interface.

Toric AI

Simplifying Data Pipeline ETL workflows with AI for the Construction Industry

Prototypes used for the Toric AI launch event

Created and prototyped by Renato Ferro for internal and marketing purposes

What

Role

Lead Product Designer
Strategic Contributor
Researcher

Responsibilities

Research
User Experience Design
Visual & Interaction Design
UX Architecture
System Design
Prototyping & Iteration
Team Enablement
Design System

Team

Lead Product Designer
CEO (Acting PM)
CMO
CTO
CS Director
Senior Sales Director

Challenges

This project began before ChatGPT-3 gained widespread popularity.

Timeline: 2 months total
Undefined Vision: Initially, the lack of a clear direction integrating AI features, benefits, and costs.
Emerging Technology: Integrating LLM/RAG and other new AI technologies on an early stage posed significant UX and Dev challenges.
Limited Research and Resources: Limited time and initial customer interviews required reliance on alternative research methodologies.

Outcomes

Developed and launched three AI features, with the first being Entity Extraction for Data Quality:

Data Quality. Set of AI features that help Data Engineers to quickly normalize, classify, and extract data (entity extraction) by assessing quality and detecting errors, empowering even non-experts to boost data integrity.
Data Analysis. AI Features that automates ETL transformations driven by the users desired outcomes, enabling easy and fast insights for customers and freeing up data engineers for complex tasks.
Data Insights. Conversational experience that let users instantly query complex construction project data using natural language, eliminating the need for IT or data analysts.

Overview
Discover
Design
Develop