Tilt is a financial technology company on a mission to expand reliable access to fair cash and credit. Formerly known as Empower, the company recently rebranded to better reflect its vision of building a credit system for long-term impact, not short-term fixes.

As Tilt’s customer base grew, so did their support backlog. To scale without adding headcount, they partnered with Ada. They first deployed a generative AI agent to handle the rising volume of email inquiries, delivering faster, more accurate support without increasing strain on the team. What began as a channel-specific solution quickly evolved into a broader strategy: using AI to elevate the support experience across Tilt’s most critical touchpoints.

tackling staffing challenges with an AI agent

Tilt’s support team handles roughly 30,000 emails per month. Like many fast-growing customer service organizations, they’ve faced high turnover, making hiring and training a challenge.

To reduce that load, Tilt joined Ada’s early access program and deployed the AI agent to assist with email support. While they expected it to help, the results surpassed expectations: 64% of email inquiries were resolved by the AI agent, with no human intervention required.

Staffing is our #1 risk factor in customer service,” explained Erich Rothaus, Senior Manager, Customer Support Operations at Tilt. “Hiring an AI agent to answer customer emails means we have a resource that can immediately become an expert in our product, and is guaranteed to stay with us forever. That's incredibly valuable to our support team—and our customers, too."

The AI agent now handles routine inquiries with speed and accuracy, freeing up Tilt’s customer support team to focus on high-value interactions, building customer relationships, and providing insights that will inform important business decisions.

putting AI where it has the most context and impact

Following the success of the email pilot program, Tilt extended their AI agent to chat. Delan Diaz, the company’s CX Generative AI Manager, explains, “We have the most opportunity anywhere the user is actually authenticated. So, we moved the AI agent over to focus on chat.”

In chat, customers must be logged in, giving the AI agent access to richer, account-level context through various APIs. This unlocks the ability to handle higher-value inquiries and deliver more personalized service.

To enhance this further, Diaz’s team developed a greeting API that pulls high-level user information at the start of the conversation. Based on that data, the AI agent can dynamically filter which playbooks, processes, and knowledge to surface, making support more targeted and preemptive from the very first message.

Diaz explains, “Now when a user comes in, we can say, ‘Hey [name], we notice you have an upcoming payment on [date] for [amount] with [product], is that what you need help with?’”

Ada’s AI agent currently handles 100% of Tilt’s chat volume, and they plan to further expand coverage to their voice channel.

building confidence and unlocking performance

In financial services, support must be fast, accurate, and consistent. When Tilt first launched Ada, they were curious to see how an AI agent would compare to their human team—and how quickly it could match, or even exceed, their standards for quality.

That trust has turned into clear performance gains. Since launching Ada, Tilt’s AI agent has not only matched the accuracy of human agents but is now exceeding it in key areas:

• CSAT rose 8 points, and in some cases by as much as 15
• Containment reached 94%
• Automated resolution rates continued to climb

Diaz shares these metrics with leadership, including the CFO and the CTO, during bi-weekly and bi-monthly meetings. He attributes the measurable performance improvements to Ada’s Coaching and Playbooks features, which allow the AI agent to become more dynamic and responsive rather than follow a rigid, predefined flow. The ability to easily train the AI agent also gives Tilt more control over the customer experience.

turning chat insights into product improvements

Generative AI lets customers describe their issues in their own words and the AI agent can still resolve them. If someone asks about their credit card but really means debit, the AI can interpret the intent and respond correctly, something a scripted bot couldn’t do.

This off-script flexibility also gives the Tilt team more useful conversation data and customer insights. “With a declarative bot, it’s a scripted environment, so there’s only so much you can pull from that,” Diaz says. “When the user is telling you verbatim what they’re experiencing and you’re seeing it in their own words, the content is so much richer.”

Diaz partners with the product team to share conversational data, including analyzing transcripts to see if AI agent conversations are matching the trends they’re seeing elsewhere. This becomes even more important as more conversations are contained and don’t go into ticketing.

By going straight to the source, the Tilt team is able to identify areas of opportunity faster and more accurately, and align their product goals with what's actually happening on their customer support channels.

“Our product team has been very receptive, very eager to get more data,” Diaz explains. “Sometimes they’ll tell me what they’re working on for the upcoming quarter and ask if we have any data related to that issue, and I’ll work with our team at Ada to pull relevant transcripts.”

Rich conversational data helps the product team both validate what they’re working on and make more strategic decisions about future opportunities for improvement.

looking ahead: putting AI on the frontline of every channel

As an AI-forward company, Tilt is excited to expand their use of Ada across more channels and deepen how the AI agent supports their customers. Their innovation and relentless focus on the customer experience earned them the Most Skilled AI Agent award at Ada Interact . Their current focus is on expanding their team to better support the AI agent, and scaling up service with Ada.

With the rebrand, Tilt is looking to create a warmer, more welcoming feel in their customer service. “Our tone has been very neutral,” Diaz says, “and we want to be more approachable, intelligent, and soulful in the way we respond.” His team is currently working on training the AI agent to adopt this.

The next step is to add soft authentication or verification for email, so they can scale up the AI agent on that channel. Diaz says this would be a game changer—it would allow Ada to handle a huge volume of inquiries. Voice is also on the horizon as Tilt continues to expand Ada’s reach.

“Once we get that authentication piece in place, we want every email that comes in to go through Ada, the same as it does with chat,” he says. “That will allow us to scale across all of our channels, including voice with the new voice architecture.”

With access to more customer data through APIs, Diaz’s team plans to take proactive support even further, surfacing app issues to the AI agent before the customer has to explain.

With Ada as their partner, this goal is entirely within reach.