AI assistant to help homeowners manage maintenance and repairs

Timeframe
2024
Project
Starling
Category
venture

Summary

Starling is a home maintentance platform with an AI assistant that aims to help new homeowners care for their home. Over the course of 18 months, my co-founder and I explored several concepts, found a design partner, launched an mvp, and built related tools to acquire users through SEO marketing. While over 600 homeowners started using Starling, we failed to find a viable business model.
The marketing website's homepage hero.
The marketing website's homepage hero.

Problem

New homeowners take on the responsibility of caring for a property for the first time in their lives. Many are unfamiliar with what that requires. Instead of doing preventative maintenance, they're left responding to emergency repairs.

Research & Initial Thesis

Before building anything, I spoke with a few dozen homeowners. They raised many issues:
  • Some felt they didn't know what maintenance work they should be doing.
  • Many expressed challenges with finding trustworthy contractors.
  • Some emptied their bank account on their down payment and didn't expect so many first year expenses.
While none regretted their purchase, almost all lamented that they missed being able to text their landlord or property manager when an issue came up. Homeowners need a single point of contact to help them care for their largest asset.
It turns out property managers for homeowners do exist. They're often called concierge services, and they exist primarily for higher income households. Could AI help reduce the cost of this concierge service, allowing us to bring a similar service to middle income households?

Prototyping with a Design Partner

We found one of these boutique concierge services that was managing a few dozen properties in the Hudson Valley. They used a task management platform to track each home's work, and communicated with clients through SMS and email. They expressed some pain points with this setup:
  • Onboarding a new home required that they record a lot of information about the home. This quickly became disorganized.
  • Texting clients was convenient, but those conversations were mixed in with personal conversations.
  • Similar to homeowners, it was difficult to find reliable contractors.
We agreed to build a platform to help manage their work. Through several conversations with this design partner, we outlined a few key features:
  • As a concierge, I need to create a new home and record information about the home in organized way so that I may reference it before each visit.
  • As a concierge, I need to track recurring and one-time tasks for each home so that I may keep up to date with each home's maintenance.
  • As a concierge, I need to be able to communicate with homeowners about their home's tasks so that I may get their input on key decisions.
The first version of the app was focused on helping our design partner manage tasks across multiple properties.
The first version of the app was focused on helping our design partner manage tasks across multiple properties.

Launching an AI assistant

That design partnership helped teach us how to operate a concierge service. The service should give clients the confidence that their home is cared for. To establish that confidence, we designed an onboarding experience that created a maintenance plan tailored to your home.
A later, simpler iteration of our homeowner onboarding.
A later, simpler iteration of our homeowner onboarding.
This onboarding also collected crucial information about the home to provide context to an AI assistant. This assistant was then available to help answer maintenance questions. With the homeowner's permission, the assistant should also be able to automate outreach to reputable nearby contractors to collect a few quotes.
The task workspace allowed the homeowner and contactor to share updates. The AI assistant was always available to privately answer homeowner questions.
The task workspace allowed the homeowner and contactor to share updates. The AI assistant was always available to privately answer homeowner questions.
To overcome the marketplace cold start problem, we made unassigned tasks anonymously available for public view in an attempt to attract service professionals to the platform.
Service professionals could browse nearby jobs and contact homeowners.
Service professionals could browse nearby jobs and contact homeowners.
We were bootstrapping the project, so we had no advertising budget. To acquire users, we built related tools that helped us build our marketing site's SEO.
The heat pump savings calculator was one of several free tools we launched to acquire users.
The heat pump savings calculator was one of several free tools we launched to acquire users.

Outcome & Learning

Over 600 homeowners joined Starling and set up their home's maintenance checklist. We continued to improve the product experience, but it became increasingly clear that we could not deliver an offering that enough homeowners would be willing to pay for. While there was potential to position the platform as a marketplace, we did not feel that it was possible to bootstrap something so directly competitive with larger marketplaces. We could have done more to validate our direction earlier on, and this project reinforced the importance of good market testing before investing too much in product.

Team

© 2025 Off-Trail Studio, LLC