PROJECT

AI chatbots for business to automate customer service

Custom software development, AI chatbot setup, Customer service automation, Lead management, Marketing automation, Multi-channel integrations, Analytics dashboards

  • Industry: Marketing services
  • Croatia Location: Croatia
  • Delivery time: 16 weeks
  • Technologies: Python, PHP
4+ months

This client’s marketing agency spends time coordinating customer support, lead handling, and marketing activity across separate channels. The solution is a business automation system for AI chatbots that takes in company knowledge, customer inquiries, lead data, and campaign data. It gives the agency chatbot conversations, qualified leads, and performance views. The system serves websites, apps, and social channels. It also supports messaging, email, and voice calls.

Summary

Problem

The agency manages support, sales and marketing activities across several channels and needs faster, more consistent responses.

Solution

The solution became a single dashboard for chatbot conversations, staff support and lead management.

Result

The software is designed to reduce manual work in customer support and marketing while maintaining an overview of leads and campaigns.

Client story

The client is a digital marketing agency from Croatia. It provides digital marketing and service solutions for its clients, including chatbots, virtual assistants and lead generation.

The agency works with websites, applications and social channels. Other touchpoints include messaging platforms, email and voice calls.

Customer inquiries, leads and campaign activity require coordination.

The requirement was simple: automate customer support, manage leads and follow-ups, support multiple communication channels, and bring reporting and analytics together in one place.

The solution was designed around four connected workspaces: chatbot setup, human handover, lead qualification, and marketing and support performance. Chatbot responses were based on FAQs, manuals and product information.

The remaining workflows were based on customer and campaign data.

The client requested

  1. Automate support with chatbots that answer FAQs and common inquiries
  2. Provide dashboards, reports, logs and performance analytics
  3. Support websites, applications, social media, messaging, email and voice communication
  4. Generate, qualify, route and follow up with leads

Why off-the-shelf software did not fit

Off-the-shelf marketing platforms focus on campaign planning, content, audience segments, engagement or marketing analytics. This agency also needs support chatbots, conversational flows, escalation and lead routing in one system. A custom solution connects these elements with the agency's own content, channels, systems and workflows.

Project overview

The solution was designed around a central chatbot workspace where the agency configures conversation flows, natural language understanding, knowledge sources and escalation rules. Rasa (an open-source framework for understanding and managing conversations) supports both text and voice communication.

For customer support, a dedicated inbox brings automated conversations and assigned work together. Support staff can see the conversation history, receive a suggested reply, edit it and escalate complex inquiries when needed.

The lead workspace captures and qualifies leads, assigns them to sales representatives and manages follow-up sequences. Connections to CRM systems, customer databases and websites keep lead activity linked to business processes.

Special attention was paid to cross-channel communication. The system connects websites, mobile applications and social media. Dialogflow (a service for building conversational interfaces) and WhatsApp Business Platform are available when listed as integrations.

The reporting workspace combines campaign activity, conversations and leads. dbt-core (a tool for preparing analytical data) and grafanalib (a library for creating Grafana dashboards) support reports and visual dashboards.

Key features delivered

  • AI chatbot configuration with natural language understanding
  • Conversation flow canvas and response editor
  • Knowledge sources from FAQs, manuals and product information
  • Support inbox with suggested AI replies and escalation
  • Lead qualification, assignment, routing and follow-up sequences
  • Chatbot and support across multiple channels
  • CRM, database, website and API integrations
  • Marketing and support performance dashboards
  • Chatbot testing and response tuning

Can be added

+ Meeting scheduling + Voice call routing + Automated email campaigns + Social media publishing + SEO workflows + Additional CRM integrations

Who else can use this solution

These businesses have a similar need to coordinate customer conversations and follow-up work, so the same solution can be adapted to them.

An e-commerce business

Task It needs automated answers, lead handling, and customer engagement across its online touchpoints.
How the solution helps The solution connects chatbot support, lead qualification, follow-ups, and performance tracking in one workspace.

A healthcare business

Task It needs to manage customer inquiries, routed support work, and communication across several channels.
How the solution helps The solution provides conversational support, escalation, ticket routing, and channel-level reporting.

A digital marketing agency

Task It needs to provide chatbot, lead-generation, and automation services for its clients.
How the solution helps The solution lets the agency configure separate conversation flows, integrations, and reports for client work.

Technology stack

Python

Conversation and analytics services

Python supports the conversation, natural language, and analytics components.

PHP

Web and chatbot services

PHP supports web application and chatbot framework components.

JavaScript / TypeScript (Node.js)

Integrations and automation

Node.js supports channel connections, APIs, browser automation, and marketing integrations.

rasa-sdk

Conversation actions

rasa-sdk supports actions for text- and voice-based conversations.

rasa

Natural language understanding

rasa manages natural language understanding and dialogue flows.

dialogflow

Conversational interface integration

dialogflow connects conversational interface work when it is required.

n8n

Workflow automation

n8n connects automated steps between services and business systems.

hubspot

CRM connection

hubspot provides a connection with HubSpot.

grafanalib

Dashboard construction

grafanalib builds dashboard definitions for performance reporting.

REST APIs

Business system connections

REST APIs connect the platform with existing websites, databases, CRMs, and business systems.

Integrations

CRM systems

The solution connects lead and customer activity with CRM systems.

Websites and landing pages

Chatbots and conversational assistants run on company websites and landing pages.

WhatsApp Business Platform

The WhatsApp connection carries customer conversations through the business messaging channel.

Email and marketing tools

Email and marketing connections support follow-ups, campaigns, and customer engagement workflows.

Business databases and APIs

APIs connect chatbot and automation workflows with customer data and existing business systems.

Grafana dashboards

Analytics data feeds reporting views for conversations, leads, campaigns, and performance metrics.

The result

Chatbot setup keeps customer answers consistent

The first workspace brings bot configuration, knowledge sources, conversation flows, and escalation rules together. The agency’s FAQs, manuals, and product information form the basis for automated answers.

The test panel helps check chatbot behavior before the flow is connected to customer channels.

How it works

  1. rasa reads conversation intents and manages dialogue between the chatbot and the customer.
  2. The intent and response editor matches customer inquiries with configured answers.
  3. The test conversation panel shows how the chatbot handles the configured flow and escalation rules.
  • Configure natural language understanding
  • Add FAQs, manuals, and product information
  • Test responses and escalation behavior
Note: Knowledge sources and escalation rules stay alongside the conversation flow, so the chatbot setup reflects the agency’s service rules.
Conversation flow and chatbot setup

Conversation flow and chatbot setup

Functions on this screen
Configure AI chatbot Natural Language Understanding Configuration Conversation Flow Design Test chatbot functionality

Human support receives the conversations that need attention

The support inbox gives agents a clear view of customer inquiries, conversation history, suggested responses, and routed work. Unresolved issues can move into human support instead of staying with the automated assistant.

Ticket and call-routing status keeps the next support action visible.

How it works

  1. The conversation list shows customer inquiries and conversation history.
  2. The support workspace prepares a suggested AI response from the conversation and FAQ responses.
  3. The escalation controls show the routed ticket or call status for human support.
  • Review customer complaints
  • Edit suggested replies
  • Track routed tickets and calls
Tip: Human escalation is part of the support flow, not a separate system.
Support inbox and human hand-off

Support inbox and human hand-off

Functions on this screen
AI Support Copilot Automate customer support

Lead follow-ups stay connected to qualification

The lead workspace turns captured conversations into organized sales work. The sales team reviews qualification answers, assigns leads, and manages follow-up sequences from the same place.

Notifications and conversation history help the team keep the next contact connected to the original inquiry.

How it works

  1. The lead table displays lead data, classifications, and assignments.
  2. Qualification fields organize customer requirements and qualification answers.
  3. The follow-up sequence sends configured messages and shows notification status.
  • Review lead conversations
  • Assign leads to sales staff
  • Track follow-up messages
Lead qualification and follow-up workspace

Lead qualification and follow-up workspace

Functions on this screen
Automate Follow-Ups Generate leads for online shop

Performance reporting brings campaigns and support together

The dashboard combines campaign performance, customer interactions, generated leads, and conversation metrics. Marketing managers and analysts can compare channel activity with lead and support information in one view.

Reports and customer insights provide a shared basis for reviewing marketing and service activity.

How it works

  1. KPI cards summarize campaign data, generated leads, and customer interactions.
  2. The campaign chart and lead funnel show performance and lead movement.
  3. The conversation metrics table applies channel filters and prepares report exports.
  • Review campaign performance
  • Compare channels
  • Export reports
Marketing and support performance dashboard

Marketing and support performance dashboard

Functions on this screen
Marketing Performance Tracking CRM Analytics Dashboards

What the solution gives

With this solution, the agency can:

  • Automate answers to FAQs and common customer inquiries
  • Keep human escalation and routed support work organized
  • Qualify, assign, and follow up with leads
  • Run chatbot support across websites, apps, social channels, messaging, email, and voice
  • Connect conversations and lead work with business systems
  • Review marketing, support, and campaign performance in shared dashboards

Could ready-made systems do the job?

Before the design, the task was compared with ready-made products by their own descriptions.

What the client needs AI marketing agents Marketing platforms HubSpot Marketing Hub, Salesforce Marketing Cloud, Klaviyo Custom solution
Automate customer support — partly
Provide analytics dashboards
Support multiple communication channels partly
Manage and follow up leads partly
Fits the client's own rules: stages, roles, reports partly partly
All steps in one place — —
No manual steps between programs partly

By the public descriptions of the products and their category.

Ready-made products cover parts of the task: marketing automation, campaign analytics, customer engagement, and some lead workflows are available. A ready-made product is enough when the agency needs only one of those parts.

This client needed its own software because support conversations, escalation, lead work, marketing activity, and reporting had to follow the agency’s rules in one place. The parts would otherwise remain spread across separate products, with some steps between them still requiring manual coordination.

In the end

In the end, the solution was made to take manual work out of the path from a customer inquiry to support, lead follow-up, and marketing reporting.

FAQ

Thinking about building a similar system? Read the answers below to see how chatbot automation, integrations, channels, and reporting can fit into one solution.