Fixli

A homeowner stares at a leaking pipe. They have no idea what it costs, who to trust, or whether they're about to be overcharged. The information to answer all three questions exists — scattered across platforms none of them trust.

No items found.

Co-founder

Design Lead

Founder

User-side shipped Dec. 2025

Pro-side In active development

Overview

Fixli is anAI-powered home repair marketplace that connects homeowners and businesses with trusted local handymen through a suite of intelligent agents. The founding thesis: the home repair industry's information asymmetry is the root cause of distrust, price confusion, and delayed repairs. Six coordinated AI agents collapse that asymmetry — automating the information-gathering that currently happens through repeated, frustrating phone calls, and surfacing the pricing transparency that makes it possible to book with confidence.

I co-founded Fixli and lead all product design and brand work. The central design challenge:how do you orchestrate six distinct AI agents into an experience that feels coherent and simple rather than technically complex?

‍

Founding Insight: Why Existing Platforms Fail

Existing home repair platforms surface providers but not prices. The homeowner submits are quest, waits for quotes that may or may not arrive, evaluates them without any price anchor, and makes a high-stakes decision with no way to assess fairness. The information asymmetry is unresolved at the moment it matters most.

What made Fixli's founding thesis specific: the friction in home repair isn't primarily a discovery problem — there are plenty of ways to find a handyman. It's a conversation and coordination problem. Homeowners and service providers spend hours on repetitive back-and-forth that AI could handle in seconds: clarifying the problem, collecting job details, comparing pricing, scheduling the appointment, following up. If AI handles the repeated dialogue, both parties arrive at the actual work faster and with better information.

‍

The Six-Agent Architecture: Design for Coordination

Fixli's productis built around six AI agents, each handling a discrete part of thehomeowner-to-professional workflow. The design challenge was making thisarchitecture invisible to the user — they experience a single, coherentservice, not six separate tools.

Agent	Role in the Homeowner Journey
Chatbot Agent	Instant NLP-powered support — understands natural language problem descriptions and routes them into the right workflow
Content Recommendation	Curates personalized DIY guides and tutorials for minor repairs, empowering homeowners to handle simple tasks themselves
Task Management	Tracks ongoing and upcoming maintenance projects in one place — nothing falls through the cracks between booking and completion
Call Service	Facilitates and records communication between homeowner and pro, creating a documented trail of agreements and expectations
Quote Comparison	Displays pricing and service details from multiple pros side by side — giving homeowners the market reference they need to evaluate fairly
Task Scheduling	Manages appointment booking, availability, and reminders — ensuring both parties are aligned without manual coordination

The UX challenge across all six agents: consistency of voice and interaction pattern. Each agent has a different function, but the homeowner should experience themas one product. I designed a shared interaction grammar — how questions are phrased, how confirmations are expressed, how uncertainty is communicated —that makes the multi-agent system feel like a single assistant rather than a collection of bots.

‍

Service Design

Two-Sided Marketplace Design

Fixli serves two users with fundamentally different goals. Homeowners want clear pricing, trusted professionals, and a job completed without surprises. Providers (Pros) want qualified leads, transparent job details, and a booking process that doesn't waste their time.

The development roadmap reflects this structural reality: 1) built the homeowner-side experience first — establishing the core value proposition of AI-assisted problem description, quote comparison, and scheduling. 2) builds the provider-side — including Pro onboarding, job management, andthe tools Pros need to manage their client relationships efficiently. 3) adds content depth: DIY guides, provider project portfolios, anduser-generated job sharing.

I designed thetwo journeys to share a data model while presenting completely differentinter faces. The homeowner sees their home, their history, their upcoming jobs. The Pro sees their pipeline, their clients, their schedule. They meet at the booking layer, where both parties have enough information to commit to thework.

AI Trust Design

In high-stakes consumer decisions — a $600 plumbing repair, a $2,000 HVAC fix — AI-generated information only helps if the homeowner trusts it enough to act on it. The design problem is not accuracy alone. It is transparency: making the AI's reasoning legible at the moment the homeowner is deciding.

My design principles for Fixli's AI interactions: AI handles the repetitive and the aggregated (collecting job details, surfacing comparable prices, scheduling logistics). The homeowner retains control over the consequential (who to hire, how much to spend, when to proceed). The AI should feel like a knowledgeable assistant, not a decision-maker. Every AI output that the homeowner acts on is accompanied by the source of that information — not presented as a black-box recommendation.

‍

Brand Identity

Fixli's brand was designed to feel both competent and trustworthy — a combination the home services category largely lacks. The tagline 'Smart. Simple. Seamless.' maps directly to the product's three design commitments: intelligent automation that actually works, an interface that doesn't require a manual, and a homeowner-to-professional workflow without the usual friction points.

The brand voice principle: Fixli talks the way a knowledgeable contractor friend would talk to you. Direct, practical, not condescending. The visual system is grounded in associations with craft and competence — not the generic 'digital marketplace' blues of the category.

‍

Takeaways

Designing Fixli from a founding position has taught me that the central design challenge of multi-agent AI products is not any individual agent — it's coherence. Six capable agents that feel like six separate tools are not a product. The design work is creating the interaction grammar and information architecture that makes the whole feel simpler than the sum of its parts.

Co-founding a second time also confirmed the pattern: the founding work is always the same move. Find the information gap that creates distrust or inaction. Design the service that closes it. The domain changes. The instinct doesn't.

‍