What I did
Outcomes
Overview
In March 2024, Sentient Flow was founded with a clear goal: to develop AI Agents that communicate with believable, human-like personalities. I joined as their Founding Product Designer, when the company was just four people in a Telegram group and a backend prototype.
I led the end-to-end design of Evolve AI, a no-code Creator Suite that brings the backend’s Dynamics Engine to life through an intuitive interface. This system allowed users to build emotionally intelligent AI characters powered by sensory awareness, evolving relationships, and adaptive memory—designed for gaming, enterprise, education, and more.
I conducted in-depth user research, designed 13+ core features, built a robust component system, and helped launch the beta product that secured $200K in pre-seed funding. I continued product work post-raise, delivering two major feature updates and collaborating across design, engineering, and marketing.
Research Methods
To understand our target audience, I conducted one-on-one interviews with 7 users sourced from Discord communities tied to popular competitors like Inworld, ConvAI, and Character.ai. These participants included gamers, indie developers, and narrative designers.
I asked about their goals, frustrations, and experiences using AI character platforms. The conversations revealed a consistent desire for characters that felt emotionally believable within fictional worlds, rather than producing generic LLM outputs. Many voiced frustration with emotionally erratic responses, a lack of behavioral transparency, and overly complex tools that hindered creativity.
RPG Designer
31 / Male
“AI can be programmed to simulate the experience of nearly anyone at a D&D table—the real challenge is giving users the control to shape that nuance.”
Needs
Behavior Architect
33 / Male
“Emotional valence and arousal levels also play a role, as too much reactivity can hinder emotional openness and vulnerability.”
Needs
Unity Developer
27 / Male
“Just haven’t exposed all the necessary functions that you would think to be really useful to a developer like me. To me, it feels like a naively developed integration.”
Needs
Competitive Analysis
In parallel, I conducted deep audits of Inworld, ConvAI, and Character.ai to understand how they structured customization, memory, and in-world behavior. While each had strengths—such as rich API access or expressive dialogue—none offered a cohesive balance between emotional realism, character control, and intuitive UX. This reinforced the need for a modular, visual-first interface in Evolve that let users fine-tune everything from sentiment control to relationship dynamics—without needing to code.
Shaping the Creator Suite
I led the architecture and design of the Evolve AI Creator Suite, a no-code platform where users could craft emotionally intelligent AI characters. The product ultimately included 13 core features, anchored in five foundational categories from the backend: Characteristics, Interactions, Aspirations, Mindset, and Capabilities. These categories were featured in the Character Editor, where users could define personality traits, goals, emotional states, relationships, and other definitions. A separate Knowledgebase allowed users to upload documents and share Contextual Lore across characters. Additional modules like Analytics, Integrations, and Settings gave users visibility into Character Performance, Security Management, and Integration controls.
To support scalability and efficient collaboration with engineering, I built the first version of our component library from scratch. This included UI components such as sidebar navigation, chat inputs, modals, tooltips, tab systems, and action cards—tailored to the specific behaviors of our platform. These assets were paired with handoff documentation and detailed specs. I ran three rounds of UI QA with the frontend team to ensure implementation matched design intent and visual polish.
Preparing for Launch
In November 2024, our team pitched to investors with a mix of backend demos, Figma prototypes, and a Framer-based documentation site and marketing site I designed. These materials helped communicate our vision clearly—and contributed to securing $200K in pre-seed funding. I immediately shifted focus toward preparing a beta version, rapidly trimming the product scope to core features: the Characters Page, the five-section Character Editor, a chat modal, and a simplified Settings page. After multiple design-engineering handoffs and QA cycles, we shipped the beta product in December 2024.
Post-Launch Role Shift
After the beta release, the company scaled quickly—growing from a four-person startup to a team of fifteen. A new Head of Design joined, and I transitioned into a more collaborative role, supporting ongoing product updates and contributing to value proposition discussions. I also adapted to an agile workflow, working within Jira under the Head of Design’s direction.
The first major post-launch feature I led was the Discovery Page, which allowed users to browse and interact with AI Agents shared by others on the platform. Characters could be filtered by genre (e.g., Fantasy, Sci-Fi, History), popularity (Trending, Top, Most Active), and tags such as personality traits or voice style. Each listing included a mini character preview and a chat modal, allowing users to test interactions before diving into edits. The goal was to encourage community creativity and increase engagement across public characters.
Chat System Expansion
To deepen interaction with AI characters, I led a major expansion of the chat system across the platform. This included a full Chat Page with support for message history, pinned conversations, multimedia sharing (images, files, GIFs), and voice input/output controls. I also designed a compact mini chat modal for the Character Editor, allowing users to test character responses in real time without breaking workflow. To support this, I introduced new components for notifications, toasts, and chat settings, ensuring consistency and usability across all chat interfaces.
