Designing AI agent conversations is a key pillar to ensure effective, natural, and goal-oriented interactions.
Shaping Chatbot Conversations: How to Create Effective User Experiences with Crafter AI
In the world of conversational AI, designing conversation flows is a fundamental pillar for enabling effective, natural, and goal-oriented interactions. This was the starting point for our latest webinar, dedicated to using the Flow Designer in Crafter AI, featuring Roberto Varinelli, Data Scientist at Athics.
Why design a conversational flow?
The webinar kicked off with a humorous clip from The IT Crowd, highlighting how a single answer often isn’t enough—and how a guided conversation is key to finding the right solution. This is especially true for AI agents. Whether you’re collecting information, generating leads, or helping users make decisions, the chatbot needs to ask questions and respond based on the user’s answers.
What is a conversational flow?
As Roberto explained, a flow is a guided chat between a user and a chatbot, built with the goal of leading the user toward a specific outcome—booking a vacation, getting support, leaving contact details, and so on.
For example, during the webinar, we showed a use case where a chatbot helps a user find their ideal vacation: from requesting a pet-friendly location in Puglia to asking about available services, schedules, and payment methods.
At the heart of design: the Flow Designer
With Crafter AI’s Flow Designer, you can:
- Create structured, step-by-step conversations;
- Customize the bot’s questions and responses;
- Integrate external APIs (CRM, e-commerce, ERPs);
- Apply logic checks and conditions (e.g., verifying if an email already exists in the database);
- Automate lead generation by collecting data like name, email, phone number, and preferred time slots.
A major advantage? No technical background required. The Flow Designer is built for users without programming skills. Plus, its self-validation system helps prevent errors when building flows.
Integration with Generative AI
Another key topic was how flows interact with generative AI (LLMs). While flows are ideal for guiding users through structured and controlled journeys, generative AI shines in open-ended conversations, where users ask freeform questions and seek detailed responses.
Both tools coexist within the platform:
- LLMs for natural conversation;
- Flow Designer for structuring specific goals.
Best practices for designing flows
In the final session, Roberto shared some golden rules for creating effective flows:
- Clarity and simplicity: Flows should be straightforward and easy to follow;
- Relevance: The bot’s responses must be aligned with the user’s intent;
- Consistent tone of voice: Align the bot’s language with the brand’s personality;
- Guided options with room for personalization: Offer structured choices while allowing user input;
- Clear closure: End with follow-ups or concrete actions (e.g., connecting with a human agent).
Do flows still matter in the era of ChatGPT?
The final question may have been the most thought-provoking: “With the rise of ChatGPT, do we still need flows?”
The answer is clear: yes—more than ever. As powerful as generative AI is, it’s not enough on its own when it comes to gathering structured data, following a logical path, or integrating with external systems. Flows provide shape, structure, and purpose to interactions—and they will remain at the core of chatbot design.