LucyAI

An AI-driven growth SaaS platform utilizing end-to-end AI Agents to automate the funnel from public acquisition to private channel conversion.

website

challenge

• Logic Simplification: Turning complex AI workflows into intuitive interfaces. • User Persona: Defining target users within a fast-evolving AI market. • Branding from 0 to 1: Building a professional B2B identity from scratch. • Visual Storytelling: Translating AI capabilities into compelling pitch decks.

solution

Simplified complex AI logic by restructuring IA and user flows, while establishing a unified design system and brand identity.

impact

Achieved a 60% reduction in task path and a 15-to-1 minute drop in configuration time, while boosting daily lead processing efficiency by 40% to facilitate Pre-A funding and the successful V1.0 launch.

Redesign Logo

In Lucy's visual upgrade, we center around "Lucy," symbolizing early human exploration and transcendence through the famous Lucy fossil. This fossil represents humanity's bold, courageous, and inclusive qualities, which are reflected in the brand’s positioning.

Color Strategy: Purple remains the primary brand color, signifying nobility and futurism. For color differentiation, pink and purple are now used to distinguish the two systems, replacing the previous yellow and purple scheme. The gradient effect adds a youthful, tech-savvy touch, while magenta accents enhance brand recognition and diversity.
Shape Evolution:The old logo’s circular design was abstract and conservative, while the new logo embraces a more structured, powerful geometric form, reflecting modernity and confidence.

Brand/Page VI Finalization

Primary Brand Color: The primary color is now clearly defined as #8389FC, used for core brand elements and main interactive controls. A gradient (#8389FC → #D477E1) is applied to buttons for added depth and vibrancy.

Background Color System: New background colors (#E7E9FE, #FDF3FC) enhance spatial differentiation across page modules. Saturation is reduced to maintain contrast and readability.

Button State Optimization:

Primary Button: Three states (Normal, Hover, Press) with color variations (#8389FC → #666DFF) for user feedback.
Secondary Button: Simple gray tone to avoid dominance.
Dashed Button: Added borders and state feedback for a lightweight feel.

Interaction Experience: Color transitions in button states enhance interaction, and unified color rules reduce visual inconsistencies across components.

Information Architecture Reconstruction

The original monolithic platform caused high cognitive load and scaling bottlenecks. I decoupled the product into System 1 (AI Customer Service) and System 2 (Public Domain Acquisition), and restructured features into a clear, tree-like hierarchy. This change eliminated functional redundancy and enhanced long-term maintainability while ensuring an intuitive, scalable navigation structure for future growth.

High-level overview of the product’s functional hierarchy

Task Flow – Task Logic Reconstruction

Original Task Flow - High Cognitive Load

The original system operated "Data Collection," "Lead Analysis," and "DM Management" as three isolated modules, forcing users to repeatedly configure settings and manually navigate between pages for a single acquisition task.For non-technical SMB users, this fragmented workflow not only caused frequent cognitive interruptions but also severely contradicted their psychological expectation of an AI-driven, automated experience.

Before： The evolution of the user journey—3 main modules and 8+ sub-pages

Facing a complex structure of 3 major modules and 8+ sub-pages, users encountered extreme cognitive overload during task configuration, with a single task setup taking up to 15 minutes, leading to significant user churn.

Action

I led the architectural restructuring of an "Asynchronous Task Flow," abandoning the original "phased manual operation" model characterized by linear breakpoints. I consolidated the three isolated stages into a "Linear Guided Flow" closed-loop logic, completely eliminating navigation barriers between modules.

By introducing "Pop-up Modal Configuration," I enabled the front-loaded entry of core parameters.Users only need to complete the end-to-end setup once before task initiation, after which the system automatically enters the processing queue.

I redefined the system's smallest operational unit as an "Automated Task." By building the underlying logic of "Config-as-Task," the system is now capable of autonomously executing subsequent asynchronous processing tasks.

Furthermore, I expanded the task queue functionality to support user-defined multi-task concurrency, automated scheduling, and trigger-based execution, truly realizing an efficient automated management system from "inputting commands" to "closed-loop execution."

New Task Flow-Linear Cognitive Flow

Building on the workflow improvements, I focused on structural flattening to further reduce cognitive load and enhance asset accessibility.

after：The evolution of the user journey

I unified "Keyword Sourcing," "DM Content," and "Reply Templates" into the primary dashboard, minimizing navigation depth and boosting operational efficiency. Also, I integrated a real-time digital dashboard to provide instant monitoring of task performance and overall project progress, turning execution into a visible process.

Result

By consolidating the fragmented architecture into a centralized L1 Dashboard, the system achieved a 60% reduction in the task completion path. Core interactions were streamlined from 15 to 9 clicks, while task configuration time plummeted from 15 minutes to just 1 minute.

The new "Batch Configuration & Collective Waiting" logic eliminates the need for step-by-step manual monitoring, granting users operational freedom. Combined with a unified "Smart Acquisition" flow, the redesign boosted daily lead processing efficiency by 40% and significantly lowered the learning curve for new users