Reduction in process time
ROI within three months
Saved per month
Kind Lending’s Problem, and How Robocorp Could Help
Kind Lending is a fast growing mortgage banking startup headquartered in Santa Ana, California.
Founded in 2019, it’s rapidly expanding. Their focus is on leveraging “new technology and process efficiencies” to reinvent the mortgage lending process and introduce a contemporary approach to mortgage banking.
From the start, they knew that process automation would be essential to ensuring their success. The responsibility of selecting and implementing the right RPA solution fell onto Jacob Hightower, Automation Engineer.
Jacob had built a number of automations at his previous employer, and now once again had an opportunity to engineer and build efficient processes at Kind Lending. He knew it was of paramount importance to pick a robust solution that had the common development tools they needed, but also the flexibility and dynamism to build whatever they needed if the functionality wasn’t already available in the tool.
Already familiar with traditional RPA tools like UiPath and Blue Prism, he knew these providers not only had huge startup costs, but were also very limited in their ability to customize automations. In particular, their drag & drop interface meant it was easy to get started, but any roadblocks presented a long and arduous process. In his experience, as soon as you ran into something that wasn’t supported on their platform, their support team would have to get involved (in addition to submitting a feature enhancement request), and you would be in for a wait, simply hoping they prioritized it.
For that reason, Jacob started looking for an easy to use, open source, developer-focused RPA technology.
He knew that most automations are not so complex as to require a developer with over a decade of coding experience. On the contrary, he understood a lot are simple scripts that require knowledge of how to navigate an API, read JSON data, and enact some basic web-scraping. Simplicity was so important in this case because the majority of the employees at Kind Lending who would be building automations were not trained in software development. As a result, they needed a platform that - on the one hand - would be flexible, but on the other would still be easy enough to manipulate that staff could understand it and build their own automations.
"I expected to be able to have a platform where I could easily take an automation from concept to actual product in a short amount of time with the ability to report on the effectiveness of that automation. Robocorp definitely exceeded those expectations. The keywords and libraries already have so much functionality I need, and I very rarely have to custom code things."
- Jacob Hightower, Automation Engineer, Kindlending
Jacob’s research into open source, developer-focused RPA solutions quickly led him to Robocorp’s RPA Framework.
From there, Robocorp’s customizability, built-in reporting with logs and analytics, and friendly cost, paired with its consumption based-pricing model, worked to seal the deal.
With other RPA vendors, Kind Lending would have to buy licenses just to train their staff and have them start building automations. (At that rate, they might never see any return on that investment!) With Robocorp, however, the team could immediately try it out on their own, knowing that - worst case - if they were to fail, there would be no harm or financial loss, simply just a few hours of time wasted.
The final consideration for Jacob was scalability. Jacob looked into Robocorp’s cloud native Control Room, and immediately saw that it would provide seamless scalability, thus allowing Kind Lending to quickly create automations in tandem with business growth. Robocorp also provided Kind Lending the ability to create self-hosted environments when necessary to protect data, as it does with all customers.
Creating the Solution
In the case of Kind Lending, current implementation consists of both Attended and Unattended bots with Control Room in Robocorp Cloud, and most of these automations run on-premises.
Let’s take an example of one of these bots in action.
One of the first automations implemented at Kind Lending was put in place to help the Capital Markets team determine the fair value of loans. It had previously tackled this business process by downloading and formatting the pricing information from government loan websites. Unfortunately, this information gathering task was very tedious and time-consuming, not to mention routine. In other words? This resource-intensive, repeat exercise was a prime candidate for automation.
Previous to automation, the process used to take anywhere from 30 to 45 minutes, each and every time it was run. Now, with Robocorp’s RPA technology, it takes approximately 4 minutes. This includes the time it takes to gather all the report data, format it correctly, and input it into a shared location in which the team can either use it, or upload it into other tools. In Kind Lending’s case, they have put the bot on a schedule it can adhere to, but it’s worth noting that it also has the ability to run on demand. In either case, this 90% reduction in process time allows their Capital Markets team to dedicate their time to more interesting, revenue-generating channels.
"The automation tool has really helped in increasing the efficiency of my workflow, allowing me to simultaneously perform other functions of the process and significantly cut down on the overall processing time"
- Howard Chu, Kind Lending Wholesale Trader, Capital Markets team
From there, Jacob decided to look into using Robocorp to accelerate the quality assurance processes at Kind Lending.
For instance, whenever Kind Lending would roll out a new product, their QA team traditionally had to check the online pricing tool. This was to make sure that all rules have been correctly applied, and that it will not show products that are not available for a given scenario.
This testing often required walking through hundreds of different pricing scenarios, which led to long wait times and other distractions. As a result, the process would often take an entire day!
Now with the help of a Robocorp Assistant, QA can let the robot automatically check the online pricing tool. QA simply has to select a spreadsheet containing those different pricing scenarios, then let the bot handle all the heavy lifting. This process, from start to finish, now takes only 30 minutes. Beyond the pricing tool, the QA team also uses other automation suites built into the attendant that allow them to run full smoke tests of particular systems, in different environments, based on the options they choose and the business operation they’re fulfilling.
Fortunately, Robocorp’s built in logs make it super easy and intuitive for QA testers to see if any of the test scenarios failed. If yes, screenshots are automatically generated, which details exactly how it failed. As you may guess, this greatly reduces the time it takes to resolve issues with the online pricing tool and other processes. And if QA testers want to verify the accuracy for whatever reason? These same logs detail the logic path that the code took, so they can see exactly when, where, and how the bot detected the roadblock.
Thanks to exceptional time savings as a result of automation, the QA team’s morale has been greatly improved, and by all accounts employee satisfaction levels have risen. In particular, since the pricing tool receives regular updates, there used to be a constant pressure on the QA team to complete the testing on tight timelines. This is most easily illustrated by the fact that when tests were completed manually, some QA engineers had to stay up as late as 2am in order to complete all required testing, in advance of the next update. Since their implementation, Robocorp’s helpful bots have resolved this issue.
In the next evolution of this QA automation use case, Jacob is using a popular tool in Robocorp’s Robot Framework called “Data Driver.” As you may have already guessed, this is for conducting data-driven tests. As an example, when QA submits a spreadsheet, the automated process will split it into 10 chunks. It will then submit each chunk as a separate work item in Robocorp Control Room, and from there run a batch of robots using Robocorp’s parallel-processing feature. This makes it possible to significantly reduce execution time: instead of a single robot completing this task in 30 minutes, 10 robots can execute this process concurrently and validate all 300 different scenarios in only 3 minutes.
One of Jacob’s favorite parts of all this? To get a holistic view, he simply turns to the Control Room, which provides analytics that give a visual representation of execution status across the entire robot workforce. This is in sharp contrast to the software at his previous employer, in which Jacob had to build his own reporting and analytics (using SQL queries and Tableau business intelligence software), just to get basic info on automations such as how the automation ran, how long it ran for, where it failed, etc.
From there, another time-consuming process which was automated with Robocorp was scheduled file transfers.
The process was already partially implemented. Windows Powershell scripts were used by the operators to generate the data files, and Kind Lending then used an SFTP client manually to securely transfer them to Partner services. Because the RPA framework can run subprocesses, the PowerShell scripts were simply called by the robot, so it was not necessary to reimplement the functionality they provided. Also, there was a bonus to having all the processes in one large chain: if any step in that chain failed, the data would not be sent, and the proper parties would be alerted.
Unfortunately, this was not happening when things were being run through the scheduled jobs previously implemented. Jacob ran into a problem with the file transfer step. The SFTP servers hosted by some of the partners were not compatible with the standard openssh library calls available with the Robot Framework. Fortunately, due to the flexible nature of Robocorp’s open source platform, Jacob was able to write a simple library that spoke directly to the lower level Paramiko SSH python module. Using this library, the robot was able to successfully connect with the partners SFTP server. Problem solved.
Sometimes, creating robots to automate complex business processes can be challenging and that’s where Robocorp’s support stepped up to help.
"Robocorp's support is great! I have even seen Antti jump in to help and I cannot think of too many organizations where the founder/CEO would jump in to troubleshoot and solve customer problems. The engineering and CS teams have been super responsive with a turnaround of less than 24 hours to answer and solve complex issues! I am impressed to see how committed everyone is at Robocorp."
- Jacob Hightower, Automation Engineer, Kind Lending
Finally, let’s talk about security.
Unsurprisingly, this is of utmost importance for Kind Lending due to the nature of their business. Some data just cannot leave their environment, and Robocorp’s solution enabled them to create automations that they could run on-premise. And for non-sensitive data, such as QA testing? Kind Lending was happy to have this simply run in the Cloud, knowing that they could always use the Vault as a useful and secure feature to store user credentials.
“The main thing that was really important to us”, said Jacob, “from my direct discussions with Robocorp’s security team was the fact that their developers follow OWASP principles, and that they have SOC 2 compliance, both of which are absolutely vital”.
What Does The Future Hold For Automations at Kind Lending?
Kind Lending is currently working on automating their HR employee data information system for the employee lifecycle, and they’ve already identified additional QA and ticketing system processes to automate.
In total, they have approximately 90 to 100 business process automations that they want to implement within the loan system in the next 6 months (and many more that are currently being identified and backlogged).
It is great to see how Kind Lending is leveraging the full potential of Robocorp’s stack to transform their business and successfully reinvent the mortgage lending process with this new technology.
Although they are still early in their process automation journey, the Robocorp Assistants and Unattended automations already implemented at Kind Lending are generating savings of $5,500 per month. This is far more than the monthly subscription they pay for Robocorp.
With the automations that Kind Lending is currently implementing, Jacob expects savings of $12,000 a month before the end of the year and $50,000 a month by the end of next year. These estimates are very achievable, since the Employee Data Lifecycle automation alone is expected to save around 13 full time workers worth of work every single day.
Perhaps more important than these cost savings is the flexibility and autonomy Kind Lending now enjoys on all RPA projects they plan to implement with Robocorp.
They are assured by the fact that, if there are integrations that are not natively supported, it will be fairly straightforward to create custom libraries. And, because Robocorp is built on top of Python, the huge repository of custom integration libraries means there’s a really good chance that they already exist. And unlike RPA solutions from vendors such as UIPath and Blue Prism, by going with Robocorp, Kind Lending is not locked in. The skills learned whilst working with Robocorp are easily transferable (as opposed to gaining a deep knowledge of a proprietary RPA tool which will only be useful for as long as that tool is used).
Kind Lending’s executive team is impressed with the business value provided by Robocorp as the technology is enabling them to align with their vision.