How Kai built an AI agent-powered chatbot that seamlessly handles complex queries

We love seeing how Make’s regular community challenges inspire our talented community to build the most creative automations. That’s why this time, we challenged people to construct AI-powered customer experience solutions using Make and Voiceflow – a no-code tool for building voice and chat AI Agents.

We received many impressive submissions. However, Kai’s automation stood out. He aimed to solve a problem traditional chatbots face: they can answer basic FAQs, but they can't verify users, access real data, or trigger meaningful actions. That degrades the value for the customer.

“I wish every company that insists on having AI chatbots would implement a variation of this approach.”

Make Community Challenge jury on Kai’s solution

To address this, Kai built an intelligent customer service system based on three AI Agents that share a common knowledge base and maintain context among themselves, making the AI-powered chatbot behave much smarter and 100% consistent. The layered solution runs on a powerful, secure backend automation that creates a seamless front-end experience. 

"A lot of organizations struggle with providing a great user experience at the digital point of contact – be it a lead, a customer, or just someone seeking information. Users are scared away by bad service experiences of automated systems quite often."

Kai S. Huxmann, CEO BrainFactor GmbH, Munich/ Germany

In this story, we explain how Kai’s elegant and modular architecture removes friction from customer interactions, saves support teams countless hours, and creates trust through One-Time Password (OTP) verification.

 “Winner for me. Kai solves a clear problem that is often overlooked when implementing AI chats. The OTP verification is simple but very effective and cool to see.”

Make Community Challenge jury on Kai’s solution

The problem

Most online chat agents have a horrible reputation and lack access to data relevant to user queries. People reach out with questions, need an invoice resent, or seek exemplary service, but traditional chatbots can only answer basic FAQ-type questions and provide no meaningful assistance.

This leaves companies with a painful bottleneck – and users frustrated more often than not. To prevent it, every customer inquiry requires human intervention, drowning support teams in repetitive requests while potential leads slip through the cracks. 

Another issue is complexity. Without automation, creating a custom chatbot that seamlessly handles all the queries would have been an at least six-month-long software project with five to six-digit costs. What’s more, each update becomes an entirely new development project. 

Companies desperately needed a solution that could connect helpful conversations with an affordable, automated workflow in the background.

The solution

To create a superior user experience, Kai built a three-agent AI chatbot that feels like a continuous conversational partner. The workflow consists of three layers – Voiceflow for handling conversations, Make for automation and security, and CRM as a secure storage for customer-specific data.

Handling the conversational aspects with Voiceflow

The first layer of Kai’s solution uses Voiceflow to run three separate AI agents, taking care of leading the dialog, identifying customer intent, and understanding the context of the query. 

Thanks to these agents, the chatbot can answer a wide range of questions without any human intervention. It independently retrieves public product, service, and company information to provide the most valuable answer. 

No sensitive, user-specific data is involved or even accessible to the system at this point, protecting the customers’ privacy.

Make for automation and security 

In Kai’s solution, Make handles process routing between systems, user verification, and a decision-making process. Behind the scenes, Make orchestrates the entire workflow, running the scenario based on predefined conditions and user input.

Answering complex queries thanks to Make’s MCP Server 

Kai made heavy use of Make's MCP Server setup to smoothly integrate API requests to Airtable, giving the AI agents access to real-time customer data once security verification has been completed. 

When users are authenticated and verified, the system can fully self-service common requests, such as resending old invoices, providing account information, changing billing addresses, or resetting passwords – all from the same front-end. 

When it comes to new leads, the system qualifies them intelligently based on predefined ICP information, determines whether the lead is valid, and triggers appropriate follow-up workflows like offering to book a call with a sales rep.

And perhaps most impressive here is that any updated knowledge files in Voiceflow are instantly accessible to all three agents, ensuring high consistency across every interaction.

Unlocking user-specific data from CRM with One-Time Password 

The chatbot’s authentication process is one of the most significant technical breakthroughs in Kai’s solution. He built a One-Time Password (OTP) verification system that handles user-specific details deterministically between an Airtable CRM and Make, without passing it to LLMs before the authentication is successfully done. 

This keeps the sensitive information completely secure. 

The OTP authentication process is quick – Make generates a temporary verification code, users receive it via email, and have to confirm it with the chatbot. The system then validates the requesting user. If valid, Make retrieves the user-specific data from Airtable and can only then perform predefined actions based on it.

The results

Kai’s solution, built on Voiceflow, Make, and Airtable, makes customer interactions with chatbots dramatically more efficient and pleasant for both sides involved. 

Here’s what it brings to the table:

• Handles over 30% of support requests fully autonomously

• Verifies users securely end-to-end without human touch

• Qualifies leads faster and more consistently based on ICP

• Gives users instant answers instead of waiting for human agents

• Provides assistance with questions legacy chatbots can’t handle

• Agents maintain the context of the conversation, even when switching between them

"People tell me, ‘Wow, I could not imagine that it could be so clean. That would have taken us months traditionally, paying an upper 5-digit bill to an agency.’”

Kai S. Huxmann, CEO BrainFactor GmbH, Munich/ Germany

AI chatbot for companies that care

Our community challenge generated dozens of innovative automation use cases. Although it is hard to choose just one winner, Kai's solution really stood out for its elegant and modular best-practice architecture and real business impact.

What makes it unique is the combination of conversational intelligence, secure OTP verification, and real backend data access, all without requiring a single line of code. The system is flexible by design: swapping logic, adding new data sources, or extending the agent's knowledge base takes minutes, not weeks. 

This solution works for any business with recurring questions, subscription plans, or a steady stream of leads. It's built for companies that want to give customers a better experience without overhiring.

"The overall setup is extremely clean, easy to maintain, and rock solid in its separation of concerns between workflows and data. OTP performance is super stable and lightning fast with auto-expiring tokens after a few minutes."

Kai S. Huxmann, CEO BrainFactor GmbH, Munich/ Germany