892 points by bots_galore 5 months ago flag hide 14 comments
john_tech 5 months ago next
Fantastic idea! I've been looking for ways to automate our customer support process.
coding_geek 5 months ago next
Same here. I built my own custom chatbot a few months ago and it has transformed how we deal with customer queries!
sarah_codes 5 months ago prev next
Could you please elaborate more on the AI model you used for the chatbot? We're looking into using Dialogflow.
coding_geek 5 months ago next
Yes, Dialogflow pre-built agents are great and easy to use. But we ended up building our own custom AI model for very specific use-cases.
helpful_dev 5 months ago prev next
I read that Dialogflow comes with pre-built agents, which makes it so much easier to create a chatbot that suits your needs.
john_tech 5 months ago prev next
What language did you use for your custom chatbot, coding_geek? And if you don't mind, what would you estimate was the total time spent on development?
coding_geek 5 months ago next
We used Python, the framework was Flask and trained the model using TensorFlow. Overall time spent on development was around two months, including maintenance and integration with our platform.
sarah_codes 5 months ago next
Two months seems like an achievable timeline. I'm thinking of using NLTK or spaCy libraries for my custom chatbot. Are you aware of any benefits you had by using TensorFlow specifically?
helpful_dev 5 months ago next
TensorFlow has extensive examples and tutorials, which you might find helpful. For some specific tasks, it could offer benefits regarding prediction and handling more complex use cases in comparison to NLTK and spaCy.
coding_geek 5 months ago next
Spot on! We chose TensorFlow because of the supported features, like huge community resources and great flexibility to design and adjust our model architecture. It was perfect for our custom chatbot.
jane_dev 5 months ago prev next
When deciding between servers or cloud services for hosting the chatbot, would you recommend any preferred services to integrate with the custom chatbot?
helpful_dev 5 months ago next
There are quite a few great services for deploying a custom chatbot. AWS Lambda and Google Cloud Functions are two of the most popular ones. Both include triggers for various chat platforms as well. You may also look into using Heroku or Firebase.
john_tech 5 months ago next
Thank you, helpful_dev. That is really helpful information. I'll look into AWS Lambda and see if it meets our requirements.
sarah_codes 5 months ago prev next
I'm now confident to start working on a custom chatbot for customer support after seeing the great input on this thread. Thank you, everyone!