Inner Insight - IBM Watson vs API.AI - Google AI platform vs IBM AI APIs - Important facts to know and decide.

Major factors you would like to decide on for a AI based development are most likely to be below.
1. Easy integration.
2. Fewer API calls to reduce the time, probability of failures, ease of understanding of logs etc.
3. Not much to develop for front end, and let the platform handle some formats with minimal code.
4. Cheaper platform with reliability and security,  for personal, small scale and enterprises scales.
5. Not to involve any third party libraries and kits that are not reliable and may change anytime, making your code not work anymore.
6. Last but not the least, a fast development to launch features quickly.

Personally we researched and worked on two and would like to provide some deep insight for
IBM Watson Vs API.AI

Insight Factor	IBM Watson	API.AI
Ease of Integration of various services	All services are different and the AI expert needs to collaborate the Conversation Service, Alchemy API service.  Multiple API calls may be the made for a simple NLP flow instead of one call.  Examples of services and voice to text, text to speech, language support etc.	Easy to use and integrate, one API call with the input and get the output.
Price	Each service has separate price. So multiple API calls in one flow, may add up the price. Please check the prices for each API call before you decide.	API.AI is totally free, including the enterprise level scaling.
Graphics	To format the conversation output to a platform specific format e.g. cards, carousels etc. in FB, you might need to have some third party modules e.g. botkit for node js and coding is needed.	Most of the formats are pre integrated and even business users can do it via API.AI Gui. The few that are not there are very easy and no need to code, can be done with small tweaks.
Support	IBM provides some level of support in services	In API.AI , you will hardly feel any need for support and if needed, all flows and logs are transparent
Analytics	Can be integrated with any analytics. IBM analytics can be used as well, but have a separate price.	Has FREE inbuilt analytics in API.AI console to a certain level, analytics platform with more info in early access ( so all will be in one console), and custom analytics can also be done (this can be done for IBM Watson conversations as well).
Development time	Moderate, integration of various atomic services needed for your flow can make it a little longer. Also depending on what graphics are needed, it will impact the development times.	Very fast end to end development and configuration.
Integrated social   platforms	None, Need to use some libraries and modules to show the output of conversation in a desired format.	Google assistant(via google actions), Viber, Skype, Slack, Facebook, Kik, Telegram and many more. Small customization might be needed for complex formats
Language support	Widely spoken languages and not a rich language based support.	Almost same number of languages supported in API.AI.