Overview

Ravian is an autonomous AI platform designed to help users delegate real work through simple prompts. Unlike most AI tools that behave like chat applications, Ravian is designed as a system-level product — focused on execution, clarity, and long-term usability.

My role as Head of Product & Design involved leading Ravian end to end: product vision, UX strategy, interaction design, interface systems, motion principles, and early product direction.

The core challenge was not making AI powerful — it was making it calm, predictable, and trustworthy.

Problem Definition

Most AI products today suffer from the same issues:

• They rely heavily on chat interfaces

• They demand constant attention and supervision

• They blur thinking, planning, and execution into one surface

• They overwhelm users with explanations instead of outcomes

This creates cognitive fatigue. Users end up managing the AI rather than benefiting from it.

The real problem wasn’t intelligence.
It was interaction psychology.

Design Goal

Design an AI system that:

• Feels structured, not experimental

• Separates thinking from execution

• Communicates progress without anxiety

• Prioritizes outputs over conversation

• Can evolve toward background, system-level behavior

UX Research & Mental Models

Observed Patterns (from existing AI tools)

Through analysis of existing AI products and user behavior, a few patterns stood out:

• Users treat chat as a temporary space, not a workspace

• Long conversations reduce clarity over time

• Users remember results, not explanations

• Ambiguous system states create mistrust

Key Insight

Users don’t want to talk to AI all day.
They want to hand work off and get it back done.

This insight directly shaped Ravian’s interaction model.

Interaction Psychology Principles Applied

1. Separation Reduces Cognitive Load

Ravian separates Chat Mode and Agent Mode.

• Chat Mode → exploration, clarification, thinking

• Agent Mode → structured execution and task ownership

This aligns with how the human brain switches between thinking and doing. Mixing both in one interface leads to confusion and misuse.

2. Intent Should Be Lightweight

The prompt bar is intentionally minimal and persistent.

Psychological reasoning:

• Lower activation energy → higher usage

• Small inputs feel less committal

• Users are more willing to express intent briefly than start “a chat”

This makes interaction feel closer to a system command than a conversation.

3. Progress Without Anxiety

The status bar above the prompt communicates:

• what’s happening

• how far along it is

• whether attention is required

It avoids urgency signals (bright colors, aggressive animation).

This supports calm progress awareness, not dopamine-driven feedback loops.

4. Outputs Over Explanations

Humans trust systems that:

• produce visible results

• behave consistently

• don’t over-explain themselves

Ravian prioritizes files, reports, and artifacts as the primary interface. Explanations exist, but they’re secondary.

This shifts AI from assistant to executor.

Interface & System Design Decisions

Chat vs Agent Mode Toggle

A clear, explicit mode switch prevents accidental task execution and sets expectations early.

Users always know:

• whether they’re exploring

• or delegating responsibility

This increases confidence and reduces misuse.

Workspace-Based Execution

Agent execution happens in a dedicated workspace that shows:

• task steps

• generated files

• completion state

This mirrors familiar project and OS-level metaphors, reducing learning time.

Dedicated Connections Overlay

Integrations are handled in a separate overlay instead of being embedded everywhere.

Why:

• Keeps the core experience focused

• Treats integrations as system capabilities, not features

• Mirrors OS permission and account-management patterns

This improves trust and perceived security.

Motion & Visual Design Philosophy

• Motion is used to explain state, not decorate

• Visual hierarchy favors clarity over branding

• Neutral palette supports long sessions

• Accent colors signal interaction, not personality

The design avoids spectacle to support repeated, everyday use.

What’s Implemented vs Proposed (Transparency)

Implemented

• Chat & Agent mode separation

• Prompt-first interaction model

• Task execution views

• Connections management overlay

Proposed / Directional

• Background execution

• System-level invocation

• Fully ambient AI behavior

This case study reflects design leadership, not feature overclaiming.

Outcome (Design-Level)

Ravian establishes a clear mental model early:

• AI is something you delegate to, not converse with endlessly

• The system stays understandable even as tasks grow complex

• Users always know what’s happening and where results live

Even in its current state, the product feels structured, intentional, and trustworthy — setting a strong foundation for future system-level evolution.

Reflection

Ravian reflects my belief that the future of AI is not louder interfaces, but quieter systems. Designing for absence — knowing what not to show — became the most important part of the work.

This project strengthened my approach to product design:
clarity over cleverness, behavior over appearance, and trust over novelty.

My Role

Head of Product & Design
Product vision · UX strategy · Interaction design · Interface systems · Motion principles · Early-stage product direction