Nearly all major decision making tools that are currently used in business process automation and management are still proprietary and are offered from well established players. Lately, there is a new generation of companies in Fintech who place a lot more emphasis on exploring if open source technologies can take over this well established area where enterprise players are playing.

In this discussion, various aspects linked to adopting OSS tools for digital decision making in Fintech were talked about.

The panelists consisted of:

• Sujan Adiga, Lead Software Development Engineer at Razorpay
• Saurabh Kumar, Director Of Engineering at Capillary Technologies
• Piyush Goel, Senior Vice President - Engineering at Capillary Technologies

The conversation was led by Rishu Mehrotra, Principal - Technology Consultant at Boston Consulting Group.

Case Study 1: Razorpay Capital (lending platform) #

Problem statement: Razorpay Capital’s application journey involves multiple stages to get businesses onboarded and processed for loans. The final stage, underwriting, which is also a multi stage process in itself, was largely done offline. In order to move this process online, Razorpay decided to use a workflow engine.

Why go online: There were two reasons to take the underwriting process online-

1. Gaining visibility into what’s happening in underwriting functions. Once one has visibility, they will be able to figure out pain points and then improve upon them.
2. Underwriting is a risky process which cannot be enforced properly if it is offline.

Why use a workflow engine: Bringing underwriting online could have been done with Razorpay’s existing setup but it was a very complex workflow with different levels of approval. Hence, having a proper workflow engine would be more efficient. This could also be used to automate other parts of the application journey in the future.

Why Camunda:

1. Having visibility into the functions was a priority. Using Camunda, one can represent the entire application workflow as a Business Process Modeling Notation (BPMN) diagram and the product team will always have visibility into what is currently running and deployed. Additionally, they can also suggest and even make some minor changes. Many of the other solutions such as AWS Step Functions and Netflix Conductor were well suited for automated workflow. So if one’s use cases are mostly automated, such solutions would be better.
2. Camunda has good support for asynchronous workflows. Human based workflows also can easily be done with this platform. They have a task list like solution as well. Hence, it can be easily used to integrate with one’s internal dashboard, handle the human based task assignment, and then process it as well.
3. On Camunda, one doesn’t have to worry about previously running applications or instances on the previous version. They can easily deploy new versions and then make sure that everything is out of the box. So the platform has good support for versioning and backward compatibility.

Case Study 2: Capillary Technologies (Over-the-Air payment technology) #

Problem statement: Capillary’s new product called OTA (Over-the-Air payment) consisted of a process in which users needed to follow a certain order of steps. There were various sequences which had their own properties. So, they needed a system that would be extremely generic in nature to support the different processes.

Why use a workflow setup:

1. A workflow setup would work best as per their requirement for a generic system, since the process was ultimately a workflow.
2. Such a setup allows team members to preview a flow first before actually applying it to the production environment. Since Capillary is a multi-tenant organization, each organization may have different ways of implementing certain steps as workflows. A versioning capability allows them to test such flows before it is applied.

Why Camunda:

1. While dealing with a payment-related product, it is crucial that there is a rollback step available in case something is unsuccessful. This feature is built in Camunda.
2. Camunda’s versioning capability is quite helpful, as mentioned previously.
3. Camunda has the capability for immediate deployment, which AWS Step Functions does not have. Due to this, in Camunda one can immediately create a workflow. The same takes 15-20 minutes in AWS. Also, if one is creating a new version, they know the version ID of that. So, they can do a preview and if the new version is okay, then they can make it the default version.
4. Capillary being a multi-tenant service, there are many different kinds of workflows and there needs to be the ability to copy a particular workflow from a test flow to a live flow. All these things were achieved easily using Camunda.
5. Camunda supports synchronous request response models.
6. Compensation models are built very easily in Camunda.
7. Integrating micro services by defining the basic structure for each step is very straightforward on this platform. Since they initially developed each micro service into individual steps separately, they have the ability to combine these steps sequentially in the required ways. The services can all run in a single application. Hence, it was easy for them to have the multi tenancy part built and onboarded much easier.
8. Since they are using Camunda directly plugged in their application, they were able to scale it horizontally. There is support for long running workflows in Camunda and the ability to integrate APM services like New Relic.
9. No additional infrastructure is required for Camunda.
10. One can see the current running processes, where the processes are at, and can make modifications. The platform works like a back office where one can debug and resolve issues.
11. There is a task list which can be used for admin related workflows.

Are OSS tools like Camunda in a place where they can go head to head with specialized tools? #

Tools like Camunda are almost there. Most of the features used in these case studies are not provided by specialized tools yet. Even in the future there are lots of other use cases where solutions like Camunda will be very helpful as compared to specialized solutions. However, specialized tools can be easily used by non-tech people in organizations as well, whereas OSS tools are more tech people-friendly as of now.

How easy or difficult is it to plug something like an AI or an analytics engine into tools like Camunda? #

They can be easily plugged into Camunda’s workflow engine setup itself.

About Rootconf #

Rootconf is a platform for systems, SRE, and DevOps engineers to discuss practical issues and approaches towards building and managing reliable infrastructure. Since 2020, Rootconf has expanded into the following specialized verticals:

• Data privacy - in collaboration with Privacy Mode, Rootconf works with organizations and practitioners on discovering and showcasing data privacy practices, including data governance and data security.
• Optimizing costs of cloud infrastructure and cloud security.
• Cybersecurity, including network security, and how organizations - global and domestic - are tackling existing and emerging challenges to build for a safe, accountable and trusted internet.

Conversations around OSS BPM tools, business needs, and approaches are an ongoing topic at Rootconf. To participate, one can join our Telegram channel here.

About Hasgeek #

Hasgeek is a platform for building communities. Hasgeek believes that effective and sustainable communities are built in a modular manner, and with an underlying layer of infrastructure and services that enable communities to focus on the core of their work. Hasgeek provides this infrastructure, and the capabilities for communities to amplify their work and presence.