The questions mid-market leaders keep asking about AI, and how they shaped what we built.

Most of the leaders we talk to have already decided to use AI. Their questions begin after that decision is made. They want to know where their data goes, who can see it, and what the AI says when it doesn’t have the answer.

Those three questions shaped almost every choice we made while building OpenCX. This post walks through each one and what we did about it.

Where the data lives

Mid-market companies usually run on a variety of platforms. Customer history sits in a CRM, support tickets in a help desk, orders and invoices in a billing system, and a good share of everything else in spreadsheets. Each vendor keeps its own copy of your data in its own format, which means your company’s most useful information is scattered across tools it doesn’t fully control.

When a company adds AI on top of that, the data has to be gathered first. OpenCX is built to connect to whichever platforms and datasets a company already uses. Once the data comes in, the model inside OpenCX normalizes it, and from that point the data and the IP that comes with it belong to the client as a local asset.

We wanted ownership to be built into how the product works, so a client never has to negotiate for it later. OpenCX runs on IntiCo’s servers, and we look after the data on the client’s behalf, including helping them meet the compliance and regulatory requirements of their industry. Client data does not leak out to the wider world, which is a risk with most consumer AI models.

What the AI says when it doesn’t know

Anyone who has used a consumer chatbot has watched it answer a question with total confidence and get it wrong. In a customer conversation that mistake has a price. A wrong refund policy, a wrong delivery date or a wrong answer about a loan can cost a company a customer, and sometimes more than that.

OpenCX agents are trained on the client’s own data and answer ONLY from that data. When the answer isn’t there, the agent hands the conversation to a person on the client’s team. For example, if a customer asks about an exception to a return policy that nobody ever wrote down, the agent passes the question to someone who can make that call.

This is what we mean when we call the agents hallucination-free. They have nothing to draw on except what the client gave them, so there is nothing for them to make up.

Customers and employees need the same thing

Customer experience and employee experience usually get separate budgets, separate vendors and separate software. The work underneath looks very similar. A customer asking where an order is and an employee asking where an expense report is both want a fast, correct answer from systems that don’t talk to each other.

OpenCX runs agentic workflows for both. It also includes omnichannel chatbots that work across channels such as SMS, email, WhatsApp and voice, along with a business analytics dashboard built on the company’s combined data. This gives a company one framework, one body of data and one place to see what is happening.

How the pieces fit together

OpenCX houses the orchestrator, the AI agents, the model and the data, including access to the company’s data lakehouse. Keeping them in one place matters because each piece depends on the others. The orchestrator coordinates the agents, the agents rely on the model, and the model works from the data. If those parts come from four different vendors, every handoff between them is a place where something can break or leak.

The connection to a company’s existing platforms is built in as well. For example, a client does not have to replace their CRM or their help desk to get started. OpenCX pulls data from them and connects the agents to the tools their teams already use.

Who we built it for

We built OpenCX with mid-market companies in mind, in industries like manufacturing, financial services, retail and e-commerce, and hospitality. A company at that size has much of the same data and service demands as a large enterprise, but usually no large engineering team to build and maintain the connections between its systems. OpenCX takes that work on.

Larger organizations are welcome to talk with us as well, and we take on enterprise projects case by case.

OpenCX is AI built for businesses navigating the AI boom. We make AI for businesses that is comprehensive, powerful, and ownable. 

Written by Alfin, Director of Sales / Marketing at IntiCo AI