A financial services firm built five AI agents to sort incoming leads from a single contact form. The system now automatically validates which submissions are real, identifies existing customers, classifies inquiry types, and routes each one to the right department, replacing manual rule-based work.
The profile of this customer, they're a global provider of corporate fund and private client services, and they operate across multiple business units.
For example, they may offer payroll or they may offer legal services.
The biggest challenge that they had was their contact form is used for a number of reasons.
People trying to sell to them, people who are genuinely looking for their services, existing customers who have an issue, and people looking for jobs.
So there were all sorts of different reasons that people would contact this organization using the contact form. Because of the amount of leads that they had coming inbound, they struggled to validate those leads, and then they also struggled to identify how to route it to the relevant department. I went very quickly to AI because actually the rule sets that they have in place are relatively black and white, and it was quite a time-consuming process for an individual to actually run through. So the, the two main tools that we used are the Data Agent in HubSpot And it consists actually of 1, 2, 3, 4, 5 different agents for this process because not only did we need to identify whether or not they were a customer and whether it was a legitimate lead, we then needed to put them into the relevant segments.
And the second one is to use workflows to actually deploy the solution and then custom code within the solutions to leverage the API to to be able to search the database to enable us to find these existing customers. We found out actually halfway through this process that they had an AI governance team that needed to sign this process off, and they were happy by looking at HubSpot's description of how that data is used, that the data residency stays within HubSpot. They were able to sign that off. The successes that we've seen have done a number of things actually that led us to understand the accuracy of the AI, but they've also led us to understand the inaccuracy of some of the team decisions that have been made. So it's acting as a training mechanism for the team to be able to understand where they're making mistakes as well. And it's also helped identify other areas that the business didn't realize they had challenges. They're now considering using HubSpot as a ticketing system. So it's opened up another conversation and another module in HubSpot that we can actually start to build for them as well. So it's is developing work and streamlining their processes more.
