The Challenge

At a leading forecasting company, Customer Managers are part of the customer service team and serve as the main point of contact for retail clients. They understand the customer and the business context, but they are not data analysts or data engineers. Every month, retailers submit data for forecasting. The problem is that the data is not always clean, complete, or ready for analysis. This often leads to a ping-pong between the customer, customer manager, and data team to clarify and fix issues before forecasting can even start.

The challenge continues after the forecast is delivered. Customers naturally ask “what if?” questions: What happens if we remove an article? What if we change something? To answer these questions, the customer manager again needs to involve the data and AI team.

These repeated interactions make the process slow, difficult to scale, and highly dependent on specialized teams—turning relatively simple customer questions into time-consuming tasks.

The Solution

The solution was built around a simple goal: give customer service team more autonomy while reducing unnecessary dependency on the AI team.

• Resolve data issues earlier: Customer managers can identify and communicate data problems before they reach the data and AI team, reducing unnecessary back-and-forth with clients.
• Answer customer questions faster: Many statistical, mathematical, and “what-if” questions can be handled directly by the agent instead of being passed to technical experts.
• Speak the customer’s language: Customer managers can interact with the system using their own business terminology rather than needing to translate their questions into technical or statistical language.
• Turn data into immediate insight: Results are presented through visualizations that help customer managers quickly understand what is happening and communicate it back to the customer.

The result is a faster, more scalable forecasting process where customer managers can handle more of the day-to-day interaction themselves, while data and AI experts can focus on more complex analytical work.

What Was Built

explai’s agentic solution began taking shape in March 2025, following a two-day onsite requirement study. Through interviews with a customer manager, CTO, and CEO, and by exploring existing forecasting documents and workflows, the team identified four main pain points:

• Data-quality layer: explai’s existing ingestion capability was enhanced with an intelligent data-quality layer. The agent analyzes incoming data, generates metadata based on its content, and identifies potential issues before the data reaches the data and AI team. Each new submission can be checked until the data is ready for forecasting.
• Statistical and mathematical knowledge: explai was equipped with in-depth statistical and mathematical knowledge so the agent could handle forecasting and “what-if” questions directly.
• Customer manager language: explai’s agent was further adapted over the following five weeks to understand the customer manager’s business language and glossary, making it possible to generate advanced SQL queries from everyday business questions rather than statistical terminology.
• Business-oriented visualization: visualizations were adapted to the customer manager’s needs, making the results easier to understand and act on immediately.

Why It Works

It works because it is built around the customer manager’s real workflow, language, and needs—not technical complexity. explai supports recurring tasks and questions while adapting to each new context and keeping the customer manager in control.

This reduces the back-and-forth between customers, customer managers, and technical teams, making the process faster while allowing experts to focus on more complex work.

Path to autonomous data readiness

Today, the customer manager is still the gateway. The next step is to let the right person take responsibility directly. There could be two ways to achieve this:

• Client-side app: The client uploads the data and immediately receives data-quality insights, so they can fix issues themselves before the data reaches the forecasting team.
• Automatic connection: explai connects directly to the client’s data source, monitors the agreed submission schedule, and automatically notifies the responsible person when action is needed.

This solves a surprisingly practical problem. If the person responsible for the data is on vacation, for example, the issue should not become the customer manager’s problem. The system can notify the responsible person -or the appropriate backup—directly.

And that creates an important shift in responsibility: the customer manager no longer has to chase data. They can spend their time where they create more value: understanding the forecast, explaining the results, and helping the client make decisions.

The same principle applies after the forecast is delivered. Retail clients often need answers immediately because their next planning decision cannot wait. The future is therefore not only about making forecasting easier—it is about making the entire question → answer → decision cycle faster.