Reviews (1-8 of 8)
- It has good dialog tree management which greatly facilitates its use. Newer features such as folders have also greatly simplified reuse.
- Performance was always very good. We haven't had, in these years working with Watson Assistant, any bot performance issues themselves.
- It has a cool set of tools to make the dialogue really rich (more than just questions and answers).
- Does not present a possibility to map input text to a Watson object, which limits some use cases.
It needs a middleware that will support it in identifying temporal phrases for very specific cases.
- Watson Assistant is really good at understanding intents of questions without being told every example. With ~20 examples, we get high accuracy on user intents.
- Watson Assistant is extremely easy to program the dialog, the cloud-based workspace makes it so easy to navigate and make quick changes, we have our interns doing it.
- The team behind Watson Assistant is very active and always releasing good features.
- The Watson SDK makes it very easy to integrate your Watson Assistant workspace into your application.
- I'd like to see improved metrics on usage. I'd like to see more about what incoming texts are having difficulty being mapped to an intent.
- A lot of virtual agents have overlap in so many areas, I'd like to see improved and broader additions to the intent catalog Watson provides so I don't need to program them.
- I'd like to see Watson assist users with possible areas of intents that are being asked, but not covered in your dialog.
- It gives the necessary NLP capabilities for the bot building.
- Conversations are more engaging and the bot building processes is simple.
- It supports a lot of use cases and is adaptable and can be easily integrated with existing IT applications.
- There are some instances where IBM Watson requires some complex coding for proper integration which needs to be simplified.
- More features are needed to utilize AI capabilities to the bot building procedures.
- Gives a comparatively simple (for a chatbot/machine learning system!) means to construct a chatbot.
- Very clear framework and component sets
- Important to recognise that creating a chatbot is HARD. IBM Watson Assistant does a great job of making this a bit easier, but it's still a complex and time-consuming experience! I wonder if that can be improved somehow. For example, by applying some smart means to take smart guesses at the form of the data to be used, or a more visual (Scratch-like) interface.
- The user is a competent programmer with plenty of time to learn the framework.
- They are familiar with JSON and have the means to generate JSON structures efficiently and cleanly.
Less well suited if:
- The expectations of creating a chatbot are not well managed, this is not a "simple" task.
- You can create unlimited intent, entity and dialogue to segregate the questions in various groups, capture information and respond correctly. Example: if the visitors want to know about a service that the company offers. You can create intent with all variation of questions, capture necessary information for example industry, size, problems etc, and then respond correctly.
- You can train Watson Assistant within minutes and test the response.
- To build a powerful virtual assistant chatbot you might need to train for a longer time.
- The dashboard and the analytics can be improved to make it more user-friendly and and proactive in suggesting how you are doing so far
- Natural language processing: using intents and entities to better align user inputs to desired outputs.
- Great UI with a simple, low-barrier interface for designing conversational experiences.
- Solid documentation. Virtually all aspects of the platform are thoroughly documented, and the API documentation is especially excellent.
- None, really -- This solution exceeds our expectations!
Watson Assistant is perfect for creating any kind of conversational support experience, whether that's a simple "ask a question, get an answer" or a more in-depth guided discussion. In the HR domain, it can provide information to employees who ask questions in a regular conversational manner, and it can also coach people through decision-making processes or provide career guidance. When properly trained, it's amazing.
It's less appropriate for situations that require a human touch, and I'd also advise against using it if you don't have a person or small team fully dedicated to running it. Over the long-term, any bot's success will depend almost entirely on having someone monitor conversations, train intents on additional user utterances, adjust entities, etc.
Text to speech works in both bands, narrow and broad. It provides different audio formats support.
Some issues to find APIs to list the assistants and get the information about them such as skills, etc. Without those API alternatively used V1 APIs to get the list of workspaces (skills), but still can't get the information about which skills are for specific assistants.
- Best cloud services for AI.
- Better API reference and documentation.
- Better support on GitHub for fixing and feature.
- Cost effective model.
- Better roadmap can be done as per feedback of user or from GitHub request.
- Fiddle Example for API testing to test accuracy of some API like NLP.
- Speech-to-text stream sometimes give issues with wav format files.
Less appropriately depends on user requirement for some scenarios.
- Easy to use and learn.
- No need for any programming skills, easily understandable.
- Cost Efficient.
- Some times it takes more time to load the working environment in IBM Cloud.
- Service may be interrupted, stating "Performance and Availability Disruption".
- Dialogues sometimes take a long time to get active and may not respond, they need to be re-configured.