You need a company in the middle (like Twillio.com), I was working on this topic during the summer in NL, but we don't have Whatsapp services in our cloud.
Twillio can handle this (they also have other API driven communication services like SMS). and connect that to Watson.
So for production you need two payed services 1 whatsapp 2 f.i. Twillio for the api handling between Watson and Whatsapp.
I have a simple design somewhere in my archives so if interested send me a slack or an mail.
I have integrated WA with my Assistant chatbot via Twilio and also a search skill developed via Watson Discovery. My problem is with the quantity of messages generated and sent by the search skill. I found an option to restrict the number of responses and it is working perfectly in the web hosted chatbot (screenshot attached) but it is not working in the WA chat (screenshot attached). Does anyone have an idea what the problem might be and how to solve it?
Here, you can access information on how to get started with WA and your Slack Bot to chat with WA.You can also find information on cloning your Slack Bot repository to work with in GIT. There are guidelines on setting the proper parameters/configurations in working with WA and Slack Bot. Good-luck!I truly hope this helps.
Hi Yvonne, I'm not sure who from Ibm said we recommend integrating with slack only. In our opinion from the Watson assistant product team, you should be deploying your assistant wherever your users expect it most. It should be as convenient as possible for your end users; in this case, WhatsApp.
In addition, I think that knowledge center article is reaaaaaalllly old. I'll work on getting that updated. But we now have a built in simple ux for integrating with slack, Facebook, WhatsApp and other channels. It will be far simpler to use and maintain your assistant if you use the built in integration.
Hi Rohaan / Yvonne,
To clarify Yvonne's message, we (IBM Watson Assistant) do provide the ability to integrate Watson Assistant "out of the box" with WhatsApp (via Twilio). We also provide the capability to integrate with Slack, but they are separate channels and address different needs.
See David's update below about a forthcoming change for the issue described above.
Unfortunately, the error exists in slack too. Screenshot attached.
Hi Yvonne,I'm not sure who from IBM is stating that we feel that Watson Assistant should always be deployed on Slack. From a Product Management point of view for Assistant, we believe you should be deploying your Assistant on whatever channels your customers want to chat on. Most commonly this is going to be your website, or over the phone, but also includes messaging channels like WhatsApp, Slack, or Facebook, but should not be limited to any one. Assistant should help your customers wherever its most convenient.
In addition, the knowledge center article was very well written, but unfortunately its a bit out of date. Thanks for sharing it and I'll work on getting it updated. We now have a built in, simple to deploy Slack integration (as well as a few other channels including WhatsApp) that we now host and maintain and manage for you. You simply get your Slack credentials and copy/paste them into your Assistant integrations tab, and the rest just works. Its much more streamlined than building and maintaining your own integration even with sample Github code.
Thank you Mitch and Rohann,
For bringing this to my attention. And, I apologize to you Rohan for the error. It's a terrible mistake for builders of code because we rely upon published material to build software applications. How can we distinguish between 'what is' and 'what isn't' when the publication clearly, and unambiguously states what should be deployed with WA.
Thanks again Rohann and Mitch for bringing this to my attention. It's a learning experience for me as well. I hope the material gets pulled from the site, and updated as-soon-as-possible.
------------------------------Yvonne R. McGinnisDevOps (hopeful), Systems AdministrationObama Foundation, ChicagoChicago Cato, Illinois773-886-5579------------------------------
Thanks for reporting the problem, we are working on a fix. Due to end of year, you should expect the fix available on the first update in 2021! :)
Looking forward to it!! :)