Redesigning an AI product experience to improve usability, engagement & adoption

01 Overview

Barie AI is an independent AI product. I led it from scratch and rebuilt the whole thing. The earlier experience was dull and unengaging, didn't keep users around, wasn't properly usable, and lacked a lot of what the product needed. I owned the rebuild end-to-end - a new design system, a fresh line of features, dark mode, a mobile version, and the B2B side of the product.

Result in one line: a measurably more usable, more engaging product - drop-off down 40% and token usage up 60%. (full figures in §11)

02 Product context

Barie is a general-purpose, agentic AI product built by Programmers Force. Rather than only generating text, it's designed to do real work - running deep research and executing multi-step tasks across the apps people already use. It pairs a very large context window with live, cited sources to keep answers verifiable, and breaks complex requests into parallel subtasks to return evidence-backed results.

The product spans a broad feature set - deep research and live web search, app connectors for multi-app workflows, image, video and slide generation, a coding agent, reusable skills, and task scheduling - delivered across web and desktop (Windows and macOS), with an on-prem deployment option for enterprise clients. In independent GAIA benchmark testing in early 2026, Barie posted top-tier results across all levels, including the hardest (Level 3). Its users range from researchers and analysts to founders, marketers, developers, and small-business owners.

03 The problem

The earlier design held the product back. It looked dull, didn't engage users, and gave them little reason to stay - on top of real usability gaps and missing functionality.

• The experience felt dull and unengaging.
• It didn't keep users around - weak retention and drop-off.
• It wasn't properly usable; core actions were harder than they should be.
• It lacked a lot of the features and polish the product needed.
• No dark mode and no proper mobile experience.

Why rebuild: incremental tweaks couldn't fix structural problems. The cost of the status quo - lost users, low engagement - justified rebuilding from first principles.

04 My role & responsibilities

I led the product design end-to-end and rebuilt it from scratch - owning the structure, flows, interactions, and the full UI system.

• Built the product's design system from the ground up - defining its typography, color, and components.
• Led a whole new line of features from a design perspective.
• Designed and introduced dark mode, which the product didn't have before.
• Designed the mobile version of the product.
• Led the B2B side - the design for on-prem deployment solutions offered to clients.

I worked closely with Tech, Product, and Marketing - proposing many new feature ideas and helping Marketing shape creative campaigns and storylines.

05 Design process

I worked from the structure outward - fixing how the product was organized before refining screens.

06 Desktop app

The rebuilt desktop experience across its core surfaces - chat, deep research, documents, library, onboarding, and processing. Toggle light and dark, browse the thumbnails, or open any screen full-screen.

Light Dark

1 / 6

07 Mobile app

The product never had a real mobile experience. I designed one on the same system - carrying the core flows into a focused, on-the-go companion.

1 / 5

08 UX decisions

A few decisions that moved the needle (problem → decision → result):

09 Design system, UI & interactions

I rebuilt the interface on a design system I created from scratch - defining the typography, color, and components, then adding dark mode and a full mobile experience the product never had.

1 / 4

10 Before & after

Drag the handle to compare the old experience with the redesign.

BeforeAfter

11 Metrics & impact

The rebuild moved the numbers that mattered.

12 Key learnings

• Building a real design system first made dark mode, mobile, and a new feature line possible to ship consistently.
• Engagement followed usability - once the product felt clearer and less dull, people stayed and used it more.
• Designing across web, mobile, and B2B at once meant the system had to scale from day one.