Habitissimo Started With One AI Agent. Now a Whole Team Takes Its Calls.

About Habitissimo

In 2009, travelers could compare hotels and diners could compare restaurants, but anyone hiring a builder was on their own. Habitissimo launched that year to give people a way to compare quotes, check quality and know who to trust with their home repairs and renovations.

Today, a homeowner with a leaking pipe or a bathroom to redo can post the job on Habitissimo’s website or app for free. Professionals sign up, choose the kinds of jobs they take on and the areas they cover, and add credit to their Habitissimo account. Each time a matching request comes in, Habitissimo sends it their way, and a fee comes out of that credit. The professional then gets in touch with a quote, and the homeowner compares quotes, reviews and prices before hiring the one they trust.

Habitissimo now connects millions of households with a large network of home-repair professionals. Many of those professionals, as Gabriel Cuesta, Habitissimo’s CTO and Director of Operations, points out, don’t spend much time online. A sign-up form, a digital wallet or a forgotten password can slow them down when all they want is to get to the next job. In those moments, a quick phone call is often the fastest way to sort it out.

Where It Started

Much of Habitissimo’s day-to-day runs on phone calls: checking leads, helping professionals get set up, and answering their questions. Early on, Gabriel saw that AI agents could take on many of those calls, and he knew where to start.

Solar panel installation is one area where Habitissimo doesn’t rely only on homeowners posting jobs. The solar market is full of lead vendors, companies that collect the names and phone numbers of people interested in installing solar panels and sell those contacts to businesses like Habitissimo. Habitissimo buys some of them to bring more potential jobs to the solar installers on its platform. Each contact is a lead: a homeowner who may want panels, and a possible job for an installer.

But many of those leads turned out to be no good. Some had the wrong phone number, while others were old contacts that had already been sold to other companies. Since professionals pay for every request they receive, a bad lead meant a solar installer would end up paying for a job that didn’t exist. To weed out the bad ones, Habitissimo’s team had to call each person and confirm they were real and still interested in solar panels.

The calls were repetitive and high-volume, exactly the kind of work an AI agent could take on, so it became the job for Habitissimo’s first AI agent. Built on an early version of HappyRobot’s platform, the agent could take calls around the clock, far more than the team ever could. The real turning point came later, when the HappyRobot team showed Gabriel the full capabilities of the platform. That’s when Gabriel realized his team could build a whole set of AI agents themselves, each one designed step by step to take on a different kind of call.

What Needs a Phone Call

With the solar lead checker running, the team went looking for the next calls to hand over to AI agents. Most of those calls involved the professionals on the platform: either getting them started or finding them.

Getting started was the first hurdle. Many new professionals signed up but never made their first payment, and emails and WhatsApp messages asking why often went unanswered. Finding professionals was the second. In areas where homeowners were asking for work, there weren’t always enough professionals to take the jobs, so the team was calling local businesses one by one.

The calls coming in weren’t much easier. Professionals rang with questions about signing up or adding credit to their account. On the other side of the marketplace, every homeowner who reported a problem needed a customer support rep to call back, work out what happened and decide who should follow up. Every one of these problems needed a phone conversation, often outside office hours. Each one became a job for an AI agent.

Building on the HappyRobot Platform

Gabriel didn’t build the agents alone. Marta Mayol, Director of Customer Service, and developers Fernando Moreno Juvera and Pablo Villar built them with him on HappyRobot’s platform. Today, Gabriel sketches each new agent, and Pablo builds it out.

“What surprised me was how natural the voices sounded in Spanish,” Gabriel says. Just as striking was how easy the platform was to use. Setting up an agent took a few steps, not lines of code, so even people in operations without a technical background could follow how the agents worked and suggest changes themselves.

HappyRobot’s engineers worked alongside them. Together, the two teams built the library of help content the agents use to answer questions, and set up the WhatsApp follow-ups the agents send when someone doesn’t pick up. Whenever Habitissimo needs to integrate the agents with another system or tool, HappyRobot builds the connection. Gabriel describes the relationship as fluid: when he needs something, the team is there.

One Agent for Every Call

Today, Habitissimo runs a whole team of AI agents, each with its own job. 

The busiest is José, who now answers every call to Habitissimo’s customer service line, 24/7. Using the company’s own help content, José walks professionals through things like signing up or adding credit to their account, and opens a ticket for the team when a case needs a person. Habitissimo phased José in gradually, from 20% of calls in August 2025 to every call by October 13. Today, it resolves about 90% of requests on its own. Every call opens by telling the caller they’re speaking with an AI agent, and it hasn’t put people off. “Quite a few customers have asked in writing to be served by José,” Gabriel says. “It has infinite patience. It can put up with long explanations and complaints.”

The second is an onboarding agent. It took on a problem email and WhatsApp messages hadn’t solved: professionals who signed up for Habitissimo but never made their first payment. The agent calls each of them to ask what’s holding them back, trying up to three times over several days. If there’s still no answer, it follows up with a WhatsApp message. Running those conversations through AI analysis helped Habitissimo pinpoint where new sign-ups were getting stuck. Using what it learned, the team made changes to the platform. Those changes paid off: about 50% more new sign-ups now become paying customers.

Three more agents round out the team. A sales agent calls local businesses in areas where homeowners need work done, and it has already led to new sales. A retention agent calls professionals who have canceled their Habitissimo account, or who are using it less and may be about to, and alerts a salesperson when there’s a chance to win them back. And an incident agent calls homeowners who report a problem and sums up the conversation, so the right person can follow up.

In every case, the agents start the conversation and people step in where their judgment is needed. 

What’s Next

Gabriel compares today’s agents to junior employees on their way to mid-senior roles. Like any junior hire, they’ll earn more responsibility as they prove themselves, and that’s where Habitissimo is heading next.

The first step is letting the agents make more decisions on their own. Today, the incident agent summarizes a homeowner’s call, and the team decides who to call next. Soon, the agent will make that decision itself and leave a written draft, with the team stepping in only if a case reopens. To trust the agents with calls like that, Habitissimo is also making its quality checks more formal, with scored calls and testing.

The agents will also start acting earlier, reaching out before anyone has to call. When a professional isn’t getting enough work, José will review their subscription and expand the zones or categories they cover. The sales agent will take on more calls, reaching professionals the sales team doesn’t have time to contact.

The build is also gaining speed. The team now reuses components across agents, so each new agent starts from parts that already work. Next, Gabriel wants agents to be able to call each other’s workflows, letting one agent hand a task directly to another.

His rule for what comes next is simple: “If you do something three times, you need to automate it.”