AI Agents for Rule
Rule Communication is a Stockholm-based marketing automation company behind a platform for email and cross-channel customer communication.
After over 10 years of working together, Rule and Vilmate started a new AI project with two agents: one for business analytics and one for customer support.
The challenge
As campaign, customer, and support data grew across Rule’s systems, finding the right information quickly became more important.
Rule wanted the agents to cover two needs: faster access to business data for leadership and more efficient support handling using relevant past cases and previous customer conversations.
Our approach
Vilmate designed the agents around Rule’s workflows and the data already available across its systems.
Information exchange between the agents
Business analytics and customer support rely on different data and handle different types of requests. We kept them as separate agents while allowing them to exchange information when a request spans both areas.
Data from several systems in one analytical flow
Business questions can involve campaign activity, platform usage, conversions, payments, and support data. Vilmate brought the required sources together through Databricks so the executive agent could work across them.
Human review for support responses
For customer support, our team kept a person at the final step of the process. The agent prepares the response using previous conversations, while the support specialist remains responsible for sending it.
The solution
Vilmate built the solution around two connected agents, with Databricks handling business analytics and MCP opening access from external systems.
Business analytics in Databricks
Rule brings campaign and conversion data into Databricks, including emails sent, opens, clicks, and purchases.
Databricks Genie generates SQL queries from analytical requests, retrieves the data, and returns answers through a chat interface. Predefined dashboards cover recurring reporting.
Executive analytics agent
The executive agent gives Rule’s CEO direct access to information across several parts of the business.
For example, it can identify customers who contacted support during a selected period and combine that information with their platform usage, campaign activity, conversions, and payments.
Customer support agent
The support agent works with Rule’s support process in Intercom.
For each new ticket, it reviews previous customer conversations, looks for similar cases, and drafts a response based on earlier support answers. The draft then goes to a support specialist for sending.
The two agents also exchange information when a request spans both areas.
Model Context Protocol integration
Vilmate added Model Context Protocol (MCP) integration, giving external systems and developer code access to the agents.
The results
“We’ve been working with Vilmate for more than a year now and have tripled the number of our dedicated team since we started with them.
They’ve added quite a bit of value to our product. All the code they’ve delivered has been delivered in excellent quality… The team is also never behind schedule.”