River

Emotionally Intelligent Journaling Platform

River is a SaaS journaling platform that combines AI-assisted self-reflection with a privacy-first architecture. It helps users write, reflect, and understand their inner patterns through LLM-powered insights, BAML scripting, and a carefully designed calm interface.

Problem

Most journaling apps are either too basic or too clinical. They let users write, but they do not help them understand what their thoughts mean. At the same time, AI journaling products often feel unsafe because users are asked to share deeply personal thoughts without clear privacy guarantees.

Solution

River gives users a private space to write and reflect, using AI as a supportive layer rather than a replacement for self-awareness. The product helps users move from writing to genuine self-understanding.

My Role

I worked on River as the product designer, full-stack developer, and AI workflow designer. I defined the product concept, designed the journaling flow, built the frontend with Next.js and React, planned the privacy-first architecture, and designed AI-assisted journaling flows using BAML for structured prompt orchestration.

Technical Architecture

The frontend is built with Next.js and React, using TypeScript for type safety. The AI layer uses LLM-powered workflows with BAML scripting to make prompts structured, reusable, and easier to manage. This separation of AI behavior from the UI layer makes the reflection system easier to iterate on over time.

Key Features

AI-Assisted Journaling

Users write naturally and receive reflective responses powered by LLMs. The AI helps identify emotional patterns, summarize entries, generate follow-up prompts, and explore recurring themes. The goal is to augment self-reflection, not replace it.

BAML-Powered Prompt Workflows

I used BAML to structure AI prompts as typed, versioned functions rather than raw strings. This makes the AI behavior testable, reusable, and decoupled from the application code. Each prompt is a typed function with defined inputs and outputs, making the system far more maintainable than traditional prompt engineering approaches.

Privacy-First Design

Journaling data is deeply personal. The architecture was designed around user control, safe data handling, and privacy-conscious AI usage. The system maintains clear separation between journal content and AI processing, with future support for local-first and encrypted storage patterns.

Calm Interface

The UI is minimal, focused, and distraction-free. Instead of dashboards and complex controls, River provides a quiet writing experience that encourages honest reflection. The dark theme and typography were chosen to create a sense of safety and intimacy.

Tech Stack

Next.js, React, TypeScript, BAML, LLM-powered AI workflows, Tailwind CSS

Outcome

River became a strong product concept and technical foundation for an AI-assisted journaling SaaS. It demonstrates my ability to combine product thinking, full-stack development, AI workflow design, and privacy-focused architecture into a polished user-facing product. The project shows how I approach modern AI products: not just by adding AI features, but by designing thoughtful systems that respect user trust, emotional context, and long-term usability.