Recently, we talked to Mulesoft-CTO Uri Sarid about how his company, under the wing of Salesforce, is still independent, but at the same time the company is working together with the CRM giant to take Mulesoft to the next level. One of the ways that this is happening is by focusing more on Artificial Intelligence (AI). At Salesforce, they also call this Einstein.
Like no other, Sarid is able to explain how Mulesoft works technically, what steps the company has taken, what challenges the software faces, and what developments are really going to make a difference. That’s why a conversation with Sarid always guarantees new insights and trends in the integration market.
Mulesoft Anypoint API Community Manager
One of Mulesoft’s solutions which makes a big difference, according to Sarid, is the MuleSoft Anypoint API Community Manager. With the Anypoint Platform, companies can provide a software solution with an API. It doesn’t matter if the software solution runs in the cloud or on-premise. Even hybrid applications can be provided with an API with Anypoint.
Many companies with large software applications choose Anypoint because in some cases they have not yet developed an API. It also sometimes happens that they do have an API, but that Mulesoft can offer companies some necessary additions that they do not want to develop themselves. The focus is often on developing a good application and not on keeping an API afloat; a party like Mulesoft is much better at that. The additions that Mulesoft then offers are offering a broader API set, but also managing rights, limiting the number of requests and protection against, for example, DDoS attacks.
With the Anypoint API Community Manager, Mulesoft goes one step further. Many companies initially used Anypoint to connect external solutions to their application or to let partners do so. With the AnyPoint API Community Manager, the entire API can be made transparent to everyone with just a few simple clicks. One can choose a site template and indicate which APIs are available for which user groups. Or, in which regions certain APIs are or are not available.
This allows companies with large software solutions to use their API to attract new business and let other companies build integrations with their platform. The larger the community around an API of, for example, a large SaaS application, the more revenue and customers the application can generate.
The API Community Manager really removes barriers to integrate with a solution, so many companies see a significant acceleration in their growth after using this solution, says Sarid.
Cloud Information Model
In our general Dreamforce article of 2019, we already wrote that the Cloud Information Model was an important part of the Salesforce conference. Sarid also mentions in our conversation that this really is an important step. Sarid is a big supporter of the new standard and hopes it will get more traction worldwide. He states that standard schedules have already been drawn up for customer data, orders and various other types of data.
This not only simplifies the transfer of data according to Sarid. It will also be possible to use AI to analyse the quality of the data and determine semantics so that data that looks different but has the same meaning can be merged. This is much more difficult without the Cloud Information Model.
API’s become more robust, intelligent and less error-prone in 2020
Sarid also let his light shine on the near future. He argues that Mulesoft will take big steps in 2020 and that the introduction of Evented APIs is imminent. As a result, communication between APIs will become much more robust and less error-prone.
Currently, when an API is executed, it connects to a second API and then sends a piece of data. If the receiving API is unavailable due to a failure or overload, in many cases the data will be lost. After that, there is never again a check to see if there is anything missing. This has to be done differently, and according to Sarid, event-based APIs are the answer.
The idea is that APIs will operate less dependent on each other. With event-based APIs, everything is announced and all other systems can respond when they can. For example, if a new customer is created, the system can say there is a new customer. For the licensing system, this means that a license must be issued.
With current APIs, if the license server or API is unreachable, the whole process is likely to crash, and the customer is not created. A good company does send an error message to the developer or the DevOps team, but the process comes to a standstill.
With event-based APIs, the customer creation process continues, and it is not the developer or DevOps team that is in control, but the business user. This concerns the sales department, for example. They determine what needs to happen when a new customer is created. The process consists of granting a license, but also what should happen if this is impossible. The latter involves, for example: save the user data temporarily and try to license the user every ten minutes.
In this way, the process is automatically resumed and resolved without the involvement of a developer. The business user could even define ‘if it takes more than one hour to provide the license, give the customer a 10 percent discount to compensate’.
The most important change is that APIs become much more intelligent and events and workflows can be captured, making API communications less error-prone. At the moment, far too much goes wrong if an API is temporarily unavailable.
Eventually, we will move to the situation that APIs will no longer be pull-based but push-based. There is an announcement of a new activity, and the APIs that feel called can respond in real-time. Of course, the relations to whom the announcement should be sent are defined in advance.
In addition, the integration of APIs is becoming much more intelligent through AI and the Cloud Information Model, which means that fewer developers are needed, and business users can do this themselves.
End of Webhooks
By switching to event-based APIs, Webhooks will also disappear, and that’s a good thing. Sarid indicates in the conversation that he rarely encounters companies that have well documented all Webhooks or know what is connected to Webhooks. That is not at all convenient and offers the necessary uncertainty. By making everything truly API-driven, and preferably event-based APIs, it is easy to map out how data flows and what information goes where.
Tableau and Mulesoft are a strong combination
Sarid also shed light on the acquisition of Tableau by Salesforce. He argues that Mulesoft and Tableau are a very strong combination. Mulesoft is able to bring together all the data from a variety of data sources. Together with Tableau, Mulesoft can make that data even better and ensure that the single source of truth that Salesforce is striving for is also achieved. In the end, Tableau is also able to present it in a clear and good way like no other.
According to Sarid, Tableau is a very important asset to Salesforce. Especially to get the semantics of data right and to present it quickly, easily and in a good way. He states that Mulesoft and Tableau are already busy working together on the development of new solutions. According to him, we can expect a lot in 2020.