Is Dialogflow still relevant in the era of Large Language Models?
Till early 2023, Dialogflow was probably the best framework for developing chatbots.
Consider these three major changes after the wide adoption of Large Language Models like ChatGPT:
1 Chatbots are now being integrated with search – that is, they do not make a lot of sense as standalone apps. Google is doing the same with its Vertex AI Search.
2 Vertex AI search depends quite a lot on existing open source tools like langchain and vector databases.
This means Google is forced to encourage the use of open source tools as part of the workflow, even though it has in house paid offerings which compete with nearly all these open source tools.
3 Some of the biggest innovations in this hybrid field are being published by a company which is not even one of the tech giants
In other words, we are in a transition period from chatbots to chatbot + search hybrids, and it is quite hard to predict what new ideas will and will not be adopted.
Since documentation (and support) is far better when you have a large community of users, choosing the wrong tech stack might lead to a lot of future headaches.
So what are the takeaways?
If you are a programmer
This is a fairly easy decision, even if it involves some work.
1 Learn Python
2 Learn a Natural Language Processing library like spaCy, which will help you with both the preprocessing and postprocessing steps of prompt engineering
3 Use Jupyter Notebooks to develop and present your ideas
4 Focus on learning the underlying concepts of this hybrid field, instead of focusing on technologies provided by a specific company
If you are not a programmer
Unfortunately, the specific technology choice you make may become obsolete within a few months.
If you are certain that the Dialogflow chatbot you are building provides sufficient RoI for your business, and you don’t anticipate the need for a lot of support or documentation, it may not really matter if Dialogflow becomes obsolete.
Otherwise I recommend that you only use Dialogflow for pilot projects and for testing prototypes. (Dialogflow is still the easiest way for non-programmers to build chatbots)