Project Brief

The ask for this mock project was to create a tool that would help content creators produce better quality content with the use of new Artificial Intelligence (AI) technologies. The challenge was to design an AI-content generation tool for the end-user. The aim of this task was to make the process of generating content simple and fast. In 3 weeks' time, I was able to research, synthesize data, ideate, and rapidly prototype an AI-content generation tool that uses avatars to speak for the user based on the generated content.

Nov-Dec 2022 (3 weeks)

Competitive research
Primary & secondary research
Problem statement
Low-fidelity sketches
Mid-fidelity wireframes
High-fidelity prototypes


The idea of HAL is an interesting one. I sought to rehabilitate and rebrand this 1960s, menacing, sci-fi character into a helpful, temperate AI assistant that can help content creators generate content quickly.

Similar to the reaction one would get if you asked the average person how AI works or what it actually does, I had no concrete idea of the functionality of this technology. With that being said, there was a lot to learn from the domain research.
Primary research
Interviewed users Carla, Aregee, and Jessica prioritize:
1. Brand identity
2. Creativity
3. Confidence
My product design cohort put together a pool of content creators they had interviewed. I selected 3 user interviews, all of which alluded to brand identity and content creativity being the principal drivers behind their performance.
Secondary research
Quote from Phil:
"I do a lot on my own and it takes so much time"
While researching, I came across YouTuber and Brand Strategist, Phil Pallen. Similar to the primary users, Phil’s branding identity is central to his business and content creation. Phil does however have experience utilizing AI-driven content generation tools.


Competitive research

I sought out competitors that utilized similar AI-assistant tools. Jarvis, Synthesia, and to a lesser extent Persado, all have an avatar or robot as the face of their primary features. Similar to HAL, I wanted to know how interactive these AI tools were with the users.

Persado uses an interactive AI assistant that appears as a pop-up on the landing page.
Jarvis, an AI assistant resembles a happy, helpful mini-robot that the user can interact with.
Synthesia is the number one AI-focused avatar-based content generation tool.

Discovery insights

Throughout the discovery process, I had gained many insights into the experiences, goals, and constraints of these content creators. With these insights, I created an affinity map to try to make a connection between all the varying, yet similar experiences. From these insights, I was able to deduce a few nuggets of information.

Synthesized insights

Is an entrepreneurial, business owner, and growing online presence.

Promoting confidence, trust, and good communication helps represent and build the clients brand.

Content creation at scale can require long hours, collaboration, and understanding diverse user groups.

Have at least some or sufficient access to tools that have some form of AI-focused or AI content generation technologies.

Constraints can vary widely from lack of interaction with new AI tools, how specific you're branding most be, restrictive client requirements, and lack of creative freedom.

Why choose this method?

I decided to use an affinity map to find the commonality between the users. Based on the interviews, these users shared similar sentiments based on their experiences. I thought, better to lay out all the cards and then unite them behind shared experiences, which then would produce insights.

The Persona

With these insights, I was able to form a description of the problem that needed to be solved then construct a persona. This description or problem statement sought to keep the focus on the user at all times and guide me (the Designer) towards a feasible solution. The statement reads below;

Problem statement

Content creators whose skills are geared towards branding and digital marketing need access to AI-focused technologies that will rapidly generate content, increase diverse user engagement, and make digital social media distribution more effective so that they are not wasting time and resources.

Welcome Christina Bowman


Brand Strategist & Co-Founder of Amaros


38 years old, married, and lives in Charlotte, NC


Limited access and experience with AI-driven content tools.


A content generator that doesn’t take away from creative freedom.


Have more time to focus on people a lease on process and complexity.

Ideation at scale

During the ideation process, I had 2 divergent ideas of how to solve from the user need. I sketched a few low-fidelity designs to gauge what direction I wanted to go in. With the name HAL, there were many possibilities, so I tried to move forward with the avatar-based content generator option.

Low-fidelity sketches

Deciding to create an avatar-based content generator, I was ideating on ways to allow the user to use an avatar to speak for the brand by choosing a voice, selecting a face, and framing the output parameters.

Scaling up

During this transition, for practical reasons, I decided to prioritize spacing and eliminate some editor overlay containers for better user flow and function.

Mid-fidelity wireframes

In this stage, I began to design my first iteration of mid-fidelity wireframes that would solve for the user & stakeholder needs, build a complete user flow with clear direction and an endpoint, and display potential UI and interactivity spaces within the wireframes.

Desirability testing & visual concepts

Needing to apply a color palette and typeface, I designed 3 style tiles. Using platforms like Dribble and Unsplash, I added some visual elements that would build focus around the visual style. Not until I had the ability to do a desirability test with a few users would my bias begin to surface.

List of Adjectives

Testing 4 individuals on the style tiles, the ask was for them to provide 3 adjectives to describe what they see and provide other feedback if they had more to add.

Their desired style tile for me to move forward with was unanimous. They all selected the one tile that originally was an afterthought.

List of adjectives by the users
Style tile concept

The lonely one

One user liked the color palette but no one thought that this correlated with the concept.

My biased pick

This was the first style tile that I had designed. I originally had every intention to choose this one and use the others as backups. The users of course felt differently and won the day in the end.

User's desired pick & winner

Users’ all responded similarly that this style tile should be the visual framework to move forward with. All viewed this tile as more professional and fitting to a technical and AI-driven product.

Final designs

In this phase, I scaled up the mid-fidelity wireframes by applying the style tile visual elements and allowing for feedback for further iteration. The final high-fidelity tool was a semi-functional prototype with applied UI elements. I used the color palette, typeface, and other visual elements to add and replicate across the prototype.

The onboarding screens were designed to use plenty of white space and asks the user to answer one question at a time.
  1. User inputs tone-of-voice.
  2. User has the option to search for more options.
Step one
The user can select an avatar to speak for the brand or the business.
  1. Interactive containers with avatar options.
  2. Display completed digital content packages ready to submit.
Step two
Another feature available to the user is the ability to realign the avatar anywhere inside the video screen as well as adding text overlay around the image.
  1. Editor tools
  2. AI-generated text is spoken by the avatar
Step three
In the final step, HAL will prompt the user to set control parameters. Simply put, assessing how much autonomy HAL will be given.
  1. Users can choose the range of AI autonomy.
  2. Data visual display topical coverage scores including (keywords, tone-of-voice, and audience engagement).
Step four

View Prototype

How does this solve for the problem and the user need?

One of the needs of the user was to access a content generator that doesn’t take away from creative freedom. HAL allows the user control over the AI’s autonomy parameters. If the user feels they are losing creative freedom of the content generated, they can adjust these parameters or even input more brand-specific information for HAL to synthesize.

Why is HAL the right way to go?

I questioned moving forward with this concept initially. However, there is only one serious Avatar AI-content generation tool on the market and that’s Synthesia. I wanted to create a tool in which an AI would generate content and an avatar would speak aloud the content generated.


I sought out competitors that utilized similar AI-assistant tools. Jarvis, Synthesia, and to a lesser extent Persado, all have an avatar or robot as the face of their primary features. Similar to HAL, I wanted to know how interactive these AI tools were with the users.

Through plenty of instruction, research, ideation, trial, and error; I was able to design the avatar AI-content generator called “HAL.” While being fully aware of the past portrayals of HAL, I wanted to immediately grab the users’ attention and elicit a “what is this” or “this is interesting” response. Based on the users’ initial reactions, I think I’ve done my job.

Next steps

  • Make further iterations
  • Ideate on potential secondary functions for HAL to do.
  • Run a usability test on at least 5 users.

I am always open to new opportunities to exchange ideas, explore collaborations, and generate unique solutions.

Let's Talk!