05/03/2025
❤️ Our favourite tools for developing hashtag
The number of tools out there these days is endless, making it possible for everyone to find the perfect fit for their needs. We are always on a lookout to find the best toolset for each case, at every stage of the project, whether it’s research, design, development, or evaluation.
We thought we’d share some of our go-to tools with you - who knows, maybe they might inspire you! And if you have your own favourites, we’d love to hear about them too.
🔬 User Research
Understanding the audience is the first step in any Conversational AI project. We use Typeform surveys to gather data about user needs and behaviours. We also ask clients to fill in user persona and chatbot persona worksheets that we created in Canva to better understand their current processes and users.
💬 Conversation Design
We like to design conversations in Miro or draw.io which both are simple and reliable. We map out decision trees and visualise how the AI Assistant will interact with users. In this process we refer back to user persona and chatbot persona worksheets created during user research to ensure we’ve covered all user scenarios before jumping into development.
🧑💻 AI Chatbot Development
Once the design is ready, we build the AI Assistant using tools like Voiceflow, DialogFlow, Landbot or Manychat. These platforms allow us to quickly turn our conversation design into functional chatbots. We choose the right tool depending on the client requirements. For FAQ and simple text-based assistants we go for our own no-code solution - Chatbotly.
🔉Voice Assistants
With the growing use of voice interfaces, we also explore Vapi, Air AI and Bland. We use those tools to build voice AI agents both inbound and outbound calling.
🔗 Integrations
A key part of building smart AI Assistants is integration with other tools, such as CRMs, calendars, email automations, and custom APIs and endpoints. We use tools like Replit to easily deploy our custom endpoints and Make and Zapier for other more standard integrations. Integrating with messaging platforms is also important. We use Twilio to connect our chatbots with WhatsApp and SMS.
📈 Evaluation
Once the AI Assistant is live, we evaluate its performance using both built-in analytics dashboard from the chatbot development platforms and our custom analytics dashboard, Python scripts, Google Sheets and (in most complex cases) HumanFirst. These help us analyse user interactions, identify areas for improvement, and refine the chatbot’s responses, ensuring it stays effective and engaging over time.
What are your favourite tools? What's the newest one you tried?
----
We are ParsLabs, a multi-disciplinary Conversational AI Agency. Since 2018 we have collaborated on 89 projects all of which received five star reviews. Here we share our experience bringing AI Assistants to production.
Learn more: https://parslabs.org/
