The model is good. The interface is why users leave.

Ghost Pixel turns ambitious agent behavior into products people can understand, steer, and trust—from the first flow through production code.

Designed at

Slide 1 of 10: Hey Pepper — interactive voice states shown in the hands-free kitchen companion

Design the product around what the agent can actually do.

In the age of artificial intelligence, your product needs to be agent-native. Everything a human can do, an agent needs to do as well.

Product experience is shifting away from humans doing the extensive work. Instead, agents take the first pass.

Humans need an experience to review and manage what the agent is doing.

01

Agent-native product design

We shape the whole experience around what the agent can do and what the user needs to understand. Product structure, journeys, interaction patterns, interface states, and the moments where people delegate, review, redirect, or take back control.

02

Design engineering

We prototype in code and carry the final interface into production, so behavior, motion, responsiveness, and edge cases survive the trip from an idea to a shipped product. No translation layer between the design and the build.

The full scope of what we cover

  • 01

    Product strategy. User needs, model capabilities, and technical constraints turned into a product direction your team can act on.

  • 02

    Design system updates. Tokens, components, and docs — built in Figma, shipped as working React code.

  • 03

    User research. Interviews and usability tests that show you what users actually do, not what you assume they'll do.

  • 04

    Product flows & IA. End-to-end journeys mapped before pixels are placed, so scope stays clear and surprises stay small.

  • 05

    Interaction & visual design. Production-ready screens with states, spacing, and motion all accounted for — designed for real data, not lorem ipsum.

  • 06

    Agent & voice UX. Delegation, progress, review, interruption, uncertainty, and recovery designed for agents people can see and steer.

  • 07

    Interactive prototypes. Built in code, not slides — so stakeholders react to something real, not something imagined.

  • 08

    Content design. The words inside the product — labels, errors, empty states, and onboarding — written to match how users actually think.

Selected work

In the age of artificial intelligence, your product needs to be agent-native. Everything a human can do, an agent needs to do as well.

Product experience is shifting away from humans doing the extensive work. Instead, agents take the first pass.

Humans need an experience to review and manage what the agent is doing.

Slide 1 of 4: Hey Pepper — interactive voice states shown in the hands-free kitchen companion

Agent-native product design

We shape the whole experience around what the agent can do and what the user needs to understand. Product structure, journeys, interaction patterns, interface states, and the moments where people delegate, review, redirect, or take back control.

Slide 1 of 3: LiveKit agent observability — dynamic panel system exploration to customize your view

LiveKit — Agent Observability

Product and interaction design for a transcript-and-insights review experience, a simulations summary, and an exploratory panel system for customizing the investigation workspace.

Slide 1 of 4: Hey Pepper — interactive voice states shown in the hands-free kitchen companion

Agent-native product design

We shape the whole experience around what the agent can do and what the user needs to understand. Product structure, journeys, interaction patterns, interface states, and the moments where people delegate, review, redirect, or take back control.

How it works

We stay close from the first product decision through the final merge, so the interaction you approve is the interaction users get.

  1. Define exactly what ships.

    01

    We start inside the product, mapping user goals, journeys, system behavior, and technical constraints. A focused scoping session turns the problem into one agreed outcome before design work begins.

    Scope locked before the work starts

  2. Work in the same loop.

    02

    We join your Slack, Linear, and sprints. The work is async-first and Loom-heavy, with one working session each week when a decision is faster to make together.

    Daily context, one weekly working session

  3. Merge the finished interface.

    03

    Design and production code move together, then land directly in your repo with the decisions documented. What you approve is what ships — no interpretation gap in between.

    Receive code, not a design file.

Start with the problem. Choose the shape of the work second.

A focused audit maps the opportunity.

A project ships one defined outcome.

A monthly engagement brings design and code into the roadmap every week.

Start here

Design audit

$1,500

We spend a week inside your product, reviewing the experience from core journeys to agent behavior. You get a specific, prioritized roadmap back. Most engagements start here.

Journeys, flows, UI, and agent behavior
10–15 prioritized issues with exact fixes
30/60/90-day action plan
Written report + Loom in one week

Fixed-scope project

$6,000

We scope it with you, then design and build it directly in your repo — founder-led from kickoff to merge, so nothing gets lost along the way.

Scope locked before work starts
Design and code in one pass
Production-ready and merged into your repo
Pay only for work that ships

Bring the product problem blocking your team.

In 25 minutes, we'll decide whether an audit, a focused project, or ongoing work is the right next step.

Dylan DotoloFounder & design engineer