B2C, SaaS
Rafi - Empowering business decisions
Rafi, as an AI-powered business response solution, it is designed to empower businesses to train their own AI models tailored specifically to their unique needs and objectives.
Role:
Product Designer
Deliverables:
WebApp design
Date:
March-May 2023
CREATING AN AI BUSINESS ASSISTANT
Problem Statement
As AI increasingly becomes a favored method for information retrieval, businesses are finding it challenging to dedicate enough time and resources to customize AI systems effectively, leveraging their capabilities to meet their specific needs and objectives.
Solution
Develop a web-based AI system that:
✨ Fosters knowledge building,
✨ Enhances decision-making processes,
✨ Supports the attainment of strategic objectives across various departments within their company.
This AI web system will:
Helps businesses in optimising their operations, automating repetitive tasks, and enhancing decision-making processes, all while promoting the adoption of AI and improvement.
Design process
Research
UX Design
User Flow
Wireframe
Style Guide & Ui Element
1
2
3
4
5
Stakeholders
alignment workshop
Outcome
Value proposition
✨ Improving Decision-Making. AI suitable for all technical skill levels.
A fundamental objective of Rafi is to encourage team members to use Rafi, thus, centralising knowledge allowing employees to perform their roles more effectively.
✨ Ensure Transparency and Trust
By providing insights into how AI models make predictions and recommendations, Users can understand the results produced by AI, fostering confidence in its use and adoption.
Priorities defined
✨ Dashboard: Centralized dashboard where users can access recent interactions and see imputed documents related to the content, flag answers and in some cases access to edit mode.
✨ Conversation history: Maintain a history of user interactions with Rafi, including past queries, responses, and actions taken, to facilitate continuity and context-aware responses.
✨ Customization and personalization Provide visual feedback, such as loading indicators, animations, or progress bars, to indicate when Rafi is processing a request or retrieving information.
✨ Ability to customize Rafi's settings, preferences, and integrations to align with the organization's specific needs and workflows.
Target audience
✨ Employees that will use Rafi for day-to-day tasks and access information relevant to their roles.
✨ Admins that will manage the AI system's settings, permissions, and integrations within the organisation.
✨ Management will use Rafi for higher-level decision-making, accessing insights or reports generated by the AI system to support strategic planning.
Look and feel definition
Rafi aspired to be perceived as friendly, fostering companionship and serving as a reliable ally (user’s budy) in the workplace.
Addressing target users needs
In essence, Rafi must facilitate various tasks:
✨ Step-by-step guidance. Users should grasp Rafi's usage intuitively, without formal training.
✨ Accessibility. All users, regardless of technical expertise, should effortlessly perform tasks.
✨ Feedback loop. Rafi's users can provide feedback, review, and revise information through a peer-review system.
✨ Admin functionality. Admins should have control and access to a panel for input and model training, enabling all company members to access information.
Step by step guidance & accesibility
Step-by-step guidance
Company employees
Question Examples for UI Education.
The primary section of the interface features a search bar similar to those found in familiar search engines. Additionally, to aid users in understanding Rafi's purpose, questions related to their work-related content were included. This allows users to quickly engage with the system.
Management & Company employees
Boosting Interface Usage: Suggestions for Enhancement
Tips on how to write good questions and how to change Rafi writing style to customize answer to the user needs, so not much re-writing was needed when using the interface was needed. This was particular important for Management roles
Admins
Instructions
Brief instructions on initiating the training model were included, recognizing that company admins may possess varying levels of familiarity with AI training. The aim was to ensure clarity regarding the required steps, prioritizing clarity over a minimalist interface.
Feedback loop
Accessibility and feedback loops
Allowing users to access data that feeds the responses
Users are able to retrieve the information or datasets used by the system to generate its responses or outputs. This, allows users to validate the accuracy of the responses, comprehend the rationale behind them, and potentially utilise the data for additional analysis or decision-making purposes.
Improvement and
feedback
To maintain a clear interface and encourage user involvement in content creation, I chose to implement a binary system loop. Here, users have the option to either accept the answer and instruct the system that it was satisfactory. enhance unsatisfactory answers (not mandatory; editing window can be closed, and the feedback will be stored as 'improvement needed').
Step by step guidance & accesibility
Making sense of Admin panels for different user types
All users
Simplifying the technical complexities of training data models to enhance understanding.
Quick access for experienced users as a primary action: Users that are familiar with data model training can immediately begin creating new models.
Streamlining data model training complexities for better comprehension: Examples demonstrating how to train the chosen AI model with prepared data are provided, enabling users to become acquainted with the process independently. This entails inputting the exercise data into the model and fine-tuning its parameters to enhance performance.
Admins
Editing model as step by step process interface.
In order to facilitate the understanding of the natural language model output, users are allow to suggest related question that the consider can be related to the data input.
This user feedback can be valuable for refining the model and enhancing its understanding of user queries and data inputs.
Admins
Multiple collaborators
Several collaborators can participate in model development as part of their assigned tasks, bringing various perspectives on the data input and potential queries. Reviewing and assessing suggested questions from peers, along with offering feedback or improvement suggestions, aids in enhancing the natural language model's capabilities and effectiveness in comprehending and addressing user inquiries.
Behind the scenes
PM’s flow
First low fidelity version based on the above screens
Currently I’m open to new opportunities and projects. Feel free to reach out.
Clients
Zalando
Imminently
Lingoda
OpenSC
Marley Spoon
Softonic
Unicaja
Services
UX/UI Design
Workshops
Prototyping
Strategy
Data Visualization
Interaction design
Design Systems
B2C, SaaS
Rafi - Empowering business decisions
Rafi, as an AI-powered business response solution, it is designed to empower businesses to train their own AI models tailored specifically to their unique needs and objectives.
Role:
Product Designer
Deliverables:
WebApp design
Date:
March-May 2023
CREATING AN AI BUSINESS ASSISTANT
Problem Statement
As AI increasingly becomes a favored method for information retrieval, businesses are finding it challenging to dedicate enough time and resources to customize AI systems effectively, leveraging their capabilities to meet their specific needs and objectives.
Design Process
Workshops
UX Design
User Flow
Wireframe
1
2
3
4
Style Guide & Ui Element
5
Solution
Develop a web-based AI system that fosters knowledge building, enhances decision-making processes, and supports the attainment of strategic objectives across various departments within their company.
This AI web system will:
Helps businesses in optimising their operations, automating repetitive tasks, and enhancing decision-making processes, all while promoting the adoption of AI and improvement.
Stakeholders
alignment workshop
Outcome
Value proposition
✨ Improving Decision-Making. AI suitable for all technical skill levels.
A fundamental objective of Rafi is to encourage team members to use Rafi, thus, centralising knowledge allowing employees to perform their roles more effectively. By designing a user-friendly platform that promotes daily use, while providing support enabling everyone to add value to data file enhancements.
✨ Ensure Transparency and Trust:
Rafi prioritizes transparency and trust in AI-driven decision-making processes. By providing insights into how AI models make predictions and recommendations, Users can understand the results produced by AI, fostering confidence in its use and adoption.
Target audience
✨ Employees that will use Rafi for day-to-day tasks and access information relevant to their roles.
✨ Admins that will manage the AI system's settings, permissions, and integrations within the organisation.
✨ Management will use Rafi for higher-level decision-making, accessing insights or reports generated by the AI system to support strategic planning.
Look and feel definition
Rafi aspired to be perceived as friendly, fostering companionship and serving as a reliable ally (user’s budy) in the workplace.
Priorities defined
✨ Dashboard: Centralized dashboard where users can access recent interactions and see imputed documents related to the content, flag answers and in some cases access to edit mode.
✨ Conversation history: Maintain a history of user interactions with Rafi, including past queries, responses, and actions taken, to facilitate continuity and context-aware responses.
✨ Customization and personalization: Provide visual feedback, such as loading indicators, animations, or progress bars, to indicate when Rafi is processing a request or retrieving information.
✨ Ability to customize Rafi's settings: preferences, and integrations to align with the organization's specific needs and workflows.
Addressing target users needs
In essence, Rafi must facilitate various tasks:
✨ Step-by-step guidance. Users should grasp Rafi's usage intuitively, without formal training.
✨ Accessibility. All users, regardless of technical expertise, should effortlessly perform tasks.
✨ Feedback loop. Rafi's users can provide feedback, review, and revise information through a peer-review system.
✨ Admin functionality. Admins should have control and access to a panel for input and model training, enabling all company members to access information.
Step by step guidance & accesibility
Step-by-step guidance
Company employees
Question Examples for UI Education.
The primary section of the interface features a search bar similar to those found in familiar search engines. Additionally, to aid users in understanding Rafi's purpose, questions related to their work-related content were included. This allows users to quickly engage with the system.
Management & Company employees
Boosting Interface Usage: Suggestions for Enhancement
Tips on how to write good questions and how to change Rafi writing style to customize answer to the user needs, so not much re-writing was needed when using the interface was needed. This was particular important for Management roles
Admins
Instructions
Brief instructions on initiating the training model were included, recognizing that company admins may possess varying levels of familiarity with AI training. The aim was to ensure clarity regarding the required steps, prioritizing clarity over a minimalist interface.
Feedback loop
Accessibility and feedback loops
Allowing users to access data that feeds the responses
Users are able to retrieve the information or datasets used by the system to generate its responses or outputs. This, allows users to validate the accuracy of the responses, comprehend the rationale behind them, and potentially utilise the data for additional analysis or decision-making purposes.
Improvement and
feedback
To maintain a clear interface and encourage user involvement in content creation, I chose to implement a binary system loop. Here, users have the option to either accept the answer and instruct the system that it was satisfactory. enhance unsatisfactory answers (not mandatory; editing window can be closed, and the feedback will be stored as 'improvement needed').
Step by step guidance & accesibility
Making sense of Admin panels for different user types
All users
Simplifying the technical complexities of training data models to enhance understanding.
Quick access for experienced users as a primary action: Users that are familiar with data model training can immediately begin creating new models.
Streamlining data model training complexities for better comprehension: Examples demonstrating how to train the chosen AI model with prepared data are provided, enabling users to become acquainted with the process independently. This entails inputting the exercise data into the model and fine-tuning its parameters to enhance performance.
Admins
Editing model as step by step process interface.
In order to facilitate the understanding of the natural language model output, users are allow to suggest related question that the consider can be related to the data input.
This user feedback can be valuable for refining the model and enhancing its understanding of user queries and data inputs.
Admins
Multiple collaborators
Several collaborators can participate in model development as part of their assigned tasks, bringing various perspectives on the data input and potential queries. Reviewing and assessing suggested questions from peers, along with offering feedback or improvement suggestions, aids in enhancing the natural language model's capabilities and effectiveness in comprehending and addressing user inquiries.
PM’s flow
First low fidelity version based on the above screens
Behind the scenes
Currently I’m open to new opportunities and projects. Feel free to reach out.
Clients
Zalando
Imminently
Lingoda
OpenSC
Marley Spoon
Softonic
Unicaja
Services
UX/UI Design
Workshops
Prototyping
Strategy
Data Visualization
Interaction design
Design Systems
B2C, SaaS
Rafi - Empowering business decisions
Rafi, as an AI-powered business response solution, it is designed to empower businesses to train their own AI models tailored specifically to their unique needs and objectives.
Role:
Product Designer
Deliverables:
WebApp design
Date:
March-May 2023
CREATING AN AI BUSINESS ASSISTANT
Problem Statement
As AI increasingly becomes a favored method for information retrieval, businesses are finding it challenging to dedicate enough time and resources to customize AI systems effectively, leveraging their capabilities to meet their specific needs and objectives.
Solution
Develop a web-based AI system that fosters knowledge building, enhances decision-making processes, and supports the attainment of strategic objectives across various departments within their company.
This AI web system will:
Helps businesses in optimising their operations, automating repetitive tasks, and enhancing decision-making processes, all while promoting the adoption of AI and improvement.
Stakeholders
alignment workshop
I helped running several workshops with stakeholders and colleagues to define the platform's value proposition, outline its key features, identify the target audience, establish the desired look and feel and brand type.
Outcome
Value proposition
✨ Improving Decision-Making. AI suitable for all technical skill levels.
A fundamental objective of Rafi is to encourage team members to use Rafi, thus, centralising knowledge allowing employees to perform their roles more effectively. By designing a user-friendly platform that promotes daily use, while providing support enabling everyone to add value to data file enhancements.
✨ Ensure Transparency and Trust
Rafi prioritizes transparency and trust in AI-driven decision-making processes. By providing insights into how AI models make predictions and recommendations, Users can understand the results produced by AI, fostering confidence in its use and adoption.
Priorities defined
✨ Dashboard: Centralized dashboard where users can access recent interactions and see imputed documents related to the content, flag answers and in some cases access to edit mode.
✨ Conversation history: Maintain a history of user interactions with Rafi, including past queries, responses, and actions taken, to facilitate continuity and context-aware responses.
✨ Customization and personalization Provide visual feedback, such as loading indicators, animations, or progress bars, to indicate when Rafi is processing a request or retrieving information.
✨ Ability to customize Rafi's settings, preferences, and integrations to align with the organization's specific needs and workflows.
Look and feel definition
Rafi aspired to be perceived as friendly, fostering companionship and serving as a reliable ally (user’s budy) in the workplace.
Target audience
✨ Employees that will use Rafi for day-to-day tasks and access information relevant to their roles.
✨ Admins that will manage the AI system's settings, permissions, and integrations within the organisation.
✨ Management will use Rafi for higher-level decision-making, accessing insights or reports generated by the AI system to support strategic planning.
Addressing target users needs
In essence, Rafi must facilitate various tasks:
✨ Step-by-step guidance. Users should grasp Rafi's usage intuitively, without formal training.
✨ Accessibility. All users, regardless of technical expertise, should effortlessly perform tasks.
✨ Feedback loop. Rafi's users can provide feedback, review, and revise information through a peer-review system.
✨ Admin functionality. Admins should have control and access to a panel for input and model training, enabling all company members to access information.
Step by step guidance & accesibility
Unified interface for diverse user needs
Company employees
Question Examples for UI Education.
The primary section of the interface features a search bar similar to those found in familiar search engines. Additionally, to aid users in understanding Rafi's purpose, questions related to their work-related content were included. This allows users to quickly engage with the system.
Management & Company employees
Enhancing Interface Utilization: Practical Tips for Improvement
Tips on how to write good questions and how to change Rafi writing style to customize answer to the user needs, so not much re-writing was needed when using the interface was needed. This was particular important for Management roles
Admins
Instructions
Brief instructions on initiating the training model were included, recognizing that company admins may possess varying levels of familiarity with AI training. The aim was to ensure clarity regarding the required steps, prioritizing clarity over a minimalist interface.
Feedback loop
Accessibility and feedback loops
Allowing users to access data that feeds the responses
Users are able to retrieve the information or datasets used by the system to generate its responses or outputs. This, allows users to validate the accuracy of the responses, comprehend the rationale behind them, and potentially utilise the data for additional analysis or decision-making purposes.
Improvement and feedback
To maintain a clear interface and encourage user involvement in content creation, I chose to implement a binary system loop. Here, users have the option to either accept the answer and instruct the system that it was satisfactory. enhance unsatisfactory answers (not mandatory; editing window can be closed, and the feedback will be stored as 'improvement needed').
Step by step guidance & accesibility
Making sense of Admin panels for different user types
All users
Simplifying the technical complexities of training data models to enhance understanding.
Quick access for experienced users as a primary action: Users that are familiar with data model training can immediately begin creating new models.
Streamlining data model training complexities for better comprehension: Examples demonstrating how to train the chosen AI model with prepared data are provided, enabling users to become acquainted with the process independently. This entails inputting the exercise data into the model and fine-tuning its parameters to enhance performance.
Admins
Editing model as step by step process interface.
In order to facilitate the understanding of the natural language model output, users are allow to suggest related question that the consider can be related to the data input.
This user feedback can be valuable for refining the model and enhancing its understanding of user queries and data inputs.
Admins
Multiple collaborators
Several collaborators can participate in model development as part of their assigned tasks, bringing various perspectives on the data input and potential queries. Reviewing and assessing suggested questions from peers, along with offering feedback or improvement suggestions, aids in enhancing the natural language model's capabilities and effectiveness in comprehending and addressing user inquiries.
Behind the scenes
PM’s flow
First low fidelity version based on the above screens
Currently I’m open to new opportunities and projects. Feel free to reach out.
Currently I’m open to new opportunities and projects. Feel free to reach out.
Clients
Zalando
Imminently
Lingoda
OpenSC
Marley Spoon
Softonic
Unicaja
Clients
Zalando
Imminently
Lingoda
OpenSC
Marley Spoon
Softonic
Unicaja
Services
UX/UI Design
Workshops
Prototyping
Strategy
Data Visualization
Interaction design
Design Systems