Alfred
Alfred
Revolutionizing Asset Management with Conversational AI
Revolutionizing Asset Management with Conversational AI
Product Design
Product Design
Concept
Concept
Experience Crypto Custody as an AI Assistant
Ever find yourself frustrated with complex wallet interfaces, thinking, “I just want to send tokens to X address” or “I just want to enter a yield farming smart contract”? Meet Alfred: the solution that simplifies web3 transactions. With Alfred, you can bypass the hassle of integrating APIs and moving tokens manually. Instead, use a friendly chat interface powered by advanced LLMs. Just describe your desired transaction in natural language, and Alfred handles the rest
Project Duration
Fall 2023
Client
Neuralto Cognitive Finance
Team
Tyler Schmidt
Role
Sole Designer
Founder
Experience Crypto Custody as an AI Assistant
Ever find yourself frustrated with complex wallet interfaces, thinking, “I just want to send tokens to X address” or “I just want to enter a yield farming smart contract”? Meet Alfred: the solution that simplifies web3 transactions. With Alfred, you can bypass the hassle of integrating APIs and moving tokens manually. Instead, use a friendly chat interface powered by advanced LLMs. Just describe your desired transaction in natural language, and Alfred handles the rest
Objective
In the realm of cryptocurrency and finance, users often face overwhelming interface complexity. Transactions typically involve multiple steps due to regulatory requirements or the inherent complexity of the interactions, particularly in crypto activities such as staking and yield farming. Many users struggle to understand and engage in these processes, highlighting a need for better education and streamlined interfaces. The overall experience of financial and crypto interactions feels cumbersome, and my goal was to eliminate this interactive overhead, simplifying the process of composing and completing transactions for users.
Objective
In cryptocurrency and finance, users often struggle with complex interfaces and multi-step transactions, especially in activities like staking and yield farming. This complexity highlights the need for better education and streamlined interfaces. My goal was to simplify the process, making financial and crypto transactions more user-friendly.
Objective
In cryptocurrency and finance, users often struggle with complex interfaces and multi-step transactions, especially in activities like staking and yield farming. This complexity highlights the need for better education and streamlined interfaces. My goal was to simplify the process, making financial and crypto transactions more user-friendly.
Objective
In cryptocurrency and finance, users often struggle with complex interfaces and multi-step transactions, especially in activities like staking and yield farming. This complexity highlights the need for better education and streamlined interfaces. My goal was to simplify the process, making financial and crypto transactions more user-friendly.
Solution
Alfred simplifies crypto transactions by letting users type their desired actions into a text field. Its LLM handles the transaction composition, and conversational prompts guide users through common interactions, making complex processes and crypto management more accessible.
Solution
Alfred simplifies crypto transactions by letting users type their desired actions into a text field. Its LLM handles the transaction composition, and conversational prompts guide users through common interactions, making complex processes and crypto management more accessible.
Solution
Alfred simplifies crypto transactions by letting users type their desired actions into a text field. Its LLM handles the transaction composition, and conversational prompts guide users through common interactions, making complex processes and crypto management more accessible.
Experience Crypto Custody as an AI Assistant
Ever find yourself frustrated with complex wallet interfaces, thinking, “I just want to send tokens to X address” or “I just want to enter a yield farming smart contract”? Meet Alfred: the solution that simplifies web3 transactions. With Alfred, you can bypass the hassle of integrating APIs and moving tokens manually. Instead, use a friendly chat interface powered by advanced LLMs. Just describe your desired transaction in natural language, and Alfred handles the rest
Research
Research Goals
Our goal with the research phase was to determine the feasibility of LLM transaction Composition. Is this technically possible, and what would a product look like with this LLM driven interaction mental model. Our research was deeply informed by a year of user testing conducted for Abacus, a next-generation crypto asset management tool. The primary objective was to simplify the user experience for complex crypto interactions, enabling average users to effortlessly compose and execute advanced transactions.
Research Goals
Our goal with the research phase was to determine the feasibility of LLM transaction Composition. Is this technically possible, and what would a product look like with this LLM driven interaction mental model. Our research was deeply informed by a year of user testing conducted for Abacus, a next-generation crypto asset management tool. The primary objective was to simplify the user experience for complex crypto interactions, enabling average users to effortlessly compose and execute advanced transactions.
Methods
Our research methods consisted of rapid prototyping to generate a testable prototype soon after initial idea inception. We then presented that prototype to key stakeholders and AI engineers for feasibility testing and validation of technical capabilities. In its current phase Alfred is ready for further testing with general users.
Methods
Our research methods consisted of rapid prototyping to generate a testable prototype soon after initial idea inception. We then presented that prototype to key stakeholders and AI engineers for feasibility testing and validation of technical capabilities. In its current phase Alfred is ready for further testing with general users.
Methods
Our research methods consisted of rapid prototyping to generate a testable prototype soon after initial idea inception. We then presented that prototype to key stakeholders and AI engineers for feasibility testing and validation of technical capabilities. In its current phase Alfred is ready for further testing with general users.
Research Goals
Our goal with the research phase was to determine the feasibility of LLM transaction Composition. Is this technically possible, and what would a product look like with this LLM driven interaction mental model. Our research was deeply informed by a year of user testing conducted for Abacus, a next-generation crypto asset management tool. The primary objective was to simplify the user experience for complex crypto interactions, enabling average users to effortlessly compose and execute advanced transactions.
Methods
Our research methods consisted of rapid prototyping to generate a testable prototype soon after initial idea inception. We then presented that prototype to key stakeholders and AI engineers for feasibility testing and validation of technical capabilities. In its current phase Alfred is ready for further testing with general users.
Key Findings
Many users find the interface and processes of crypto wallets intimidating and overly complex, especially when dealing with advanced features like staking, yield farming, and multi-signature transactions.
For more detailed key findings, check out our research on Abacus.
Key Findings
Many users find the interface and processes of crypto wallets intimidating and overly complex, especially when dealing with advanced features like staking, yield farming, and multi-signature transactions.
For more detailed key findings, check out our research on Abacus.
Key Findings
Many users find the interface and processes of crypto wallets intimidating and overly complex, especially when dealing with advanced features like staking, yield farming, and multi-signature transactions.
For more detailed key findings, check out our research on Abacus.
Ideation and Concept Development
Brainstorming
Our brainstorming process for this product involved integrating our extensive knowledge of crypto wallets with the innovative potential of LLM technology and our experience with support bots. We began by conducting a user story mapping exercise to explore various directions and possibilities. This was followed by an affinity mapping exercise, which helped us identify the most resonant and impactful user stories based on our deep empathy with user needs. This collaborative and iterative approach enabled us to refine our ideas and focus on creating a solution that addresses the core challenges faced by crypto users.
Brainstorming
Our brainstorming process for this product involved integrating our extensive knowledge of crypto wallets with the innovative potential of LLM technology and our experience with support bots. We began by conducting a user story mapping exercise to explore various directions and possibilities. This was followed by an affinity mapping exercise, which helped us identify the most resonant and impactful user stories based on our deep empathy with user needs. This collaborative and iterative approach enabled us to refine our ideas and focus on creating a solution that addresses the core challenges faced by crypto users.
Brainstorming
Our brainstorming process for this product involved integrating our extensive knowledge of crypto wallets with the innovative potential of LLM technology and our experience with support bots. We began by conducting a user story mapping exercise to explore various directions and possibilities. This was followed by an affinity mapping exercise, which helped us identify the most resonant and impactful user stories based on our deep empathy with user needs. This collaborative and iterative approach enabled us to refine our ideas and focus on creating a solution that addresses the core challenges faced by crypto users.
User Story Mapping
User Story Mapping
User Story Mapping
Sketches and Early Concepts
User Flows
Our User flow design goals were to depict what the primary interaction loop of refining the users transaction preferences while using the LLM chat feature interaction or the prefabricated prompt selection to start a transaction composition.
Example of the LLM Transaction Composition Loop
User Flows
Our User flow design goals were to depict what the primary interaction loop of refining the users transaction preferences while using the LLM chat feature interaction or the prefabricated prompt selection to start a transaction composition.
Example of the LLM Transaction Composition Loop
User Flows
Our User flow design goals were to depict what the primary interaction loop of refining the users transaction preferences while using the LLM chat feature interaction or the prefabricated prompt selection to start a transaction composition.
Transaction Composition Loop
Design and Iteration
Look and Feel
With the modern suite of design tools, I prefer to compose an initial look and feel with an informal design system prior to building wireframes. This can remove some double work, and get the design docs to a flexible high fidelity quicker that still allows for flexibility through modern component systems.
The below mood board is designed to convey aspiration, and modernity. The goal is to appeal to early adopter users who may be familiar with financial apps like RobinHood.
Initial Designs
This is an example of some of the key frames in our prototype.
Working File
Early Demo Video
Reflections and Learnings
What Went Well:
This project started with strong idea inspiration. My network of collaborators was strong, and I was able to validate ideas quickly based on both engineering expertise and prior user knowledge and empathy in the space. Conversations with engineers about the feasibility of this project were extremely encouraging.
What Went Well:
This project started with strong idea inspiration. My network of collaborators was strong, and I was able to validate ideas quickly based on both engineering expertise and prior user knowledge and empathy in the space. Conversations with engineers about the feasibility of this project were extremely encouraging.
Challenges
The main challenge came from building in a bear market. When this product became a creative concept, the crypto markets were very bearish, and venture money was very tight. We decided to shelf the product for the interim after delivering a high quality concept.
Design challenges involve integrating accessibility features. We need to add a voice command interaction model to this prototype. Covering all edge cases is very challenging as a team of one.
Lessons Learned
Pursuing the integration of LLMs was well received. Conversation driven interaction can be a new standard of interaction model. Building rapidly is easy with a team of one, but its imporant to cover all edge cases.
Some times a good conversation with a business partner can lead to a valuable insight that drives a whole product concept. Its important to follow your gut and build to validate an idea.
Lessons Learned
Pursuing the integration of LLMs was well received. Conversation driven interaction can be a new standard of interaction model. Building rapidly is easy with a team of one, but its imporant to cover all edge cases.
Some times a good conversation with a business partner can lead to a valuable insight that drives a whole product concept. Its important to follow your gut and build to validate an idea.
Lessons Learned
Pursuing the integration of LLMs was well received. Conversation driven interaction can be a new standard of interaction model. Building rapidly is easy with a team of one, but its imporant to cover all edge cases.
Some times a good conversation with a business partner can lead to a valuable insight that drives a whole product concept. Its important to follow your gut and build to validate an idea.
