What is our primary use case?
Our initial use cases are mainly for finance. We are doing account payable, accounts receivable, reconciliation, and those types of things with the automation. In terms of accounts payable, we automate the invoice processing since it is an end-to-end. This means that the vendor will send an invoice to email, which will be picked up by the bot automatically. Then, it will extract the information from the invoices and post it to our SAP.
It is a web-based solution but hosted on our server.
How has it helped my organization?
The biggest driver was the cost savings. We wanted to improve productivity and save costs. Therefore, we gave most of the mundane tasks currently being done by a human to a bot. Some of the mundane tasks were reading invoices and keying in the data. We are talking about 15,000 documents every day. That is a huge volume that needs a lot of people. With the bot, it is just a fraction of the cost, because there is a huge savings in terms of manpower.
More for regulatory and audit purposes, we still require a human to approve it. Previously, we had the human to do it, then we had people cross-check it. Then, you have another layer of the approval. With the bot, we don't require two people. We only have the approval because we still have a person who does the approval, which we have to maintain.
What is most valuable?
The most important part is how easy it is to pair the automation. So, it is a canvas that is just drag and drop. You don't have to code, so it is a no-code to low-code solution.
It is good for simple tasks that we have done in the past, e.g., reading the invoices. A valuable feature is the document processing. Usually when we talk about document processing in the market, you just have OCR. Where once you extract the information, you need to program or do some type of data wrangling to actually get the value of it or process it. For Jiffy.ai, they have the machine learning behind it, so we didn't need to code one by one. For example, if you have 5,000 vendors who are sending you different types of invoices, then we are not talking about 5,000 invoices. We are talking about one vendor who has three types of templates, so that is about 15,000 documents to process. Even if you do OCR, you want to extract the information and code it to read this and that. So, Jiffy.ai has machine learning where we don't have to teach all the documents, instead we just need to teach it a few. Then, the machine will already know if it finds this type of information, then that is what it is. For example, the easiest way is the invoice number. Most vendors usually have similar wording: invoice number, invoice NO, and INV. However, in all 15,000 documents, you see that the vendors just play around with this wording. It won't differ much. Therefore, the machine learning knows because of this, you don't need to teach it all 15,000 documents. After about 10 documents, the bot can pick it up themselves and learn about it.
There are not a lot of vendors in the market who provide built-in machine learning. In the invoice, you have multiple things that you want to extract: invoice number, PO, and some other line items. With machine learning, we expect it to know what to extract from, by looking at different templates of invoices. It should know that this is similar. Even though you use the different wording across multiple templates, the machine should know that it is an invoice number. We expect the machine learning should be able to do this, and the Jiffy.ai machine learning is able to do it with 80 to 90 percent accuracy. So far, we haven't had a big problem in whatever the machine learning reads, doing it correctly. If it didn't read correctly, we would have to correct it, then the bot will learn from that, "Okay, this is actually the better way," so it can do better next time.
What needs improvement?
They are still new in the market. Or, at least, they are still a small player. They require a lot of improvement in terms of learning material as well as the community developers. If you compare Jiffy.ai to an established solution, like UiPath, you can go to YouTube and find a lot of learning material posted by UiPath, partners, and other people in the community. However, for Jiffy.ai, you won't find that available in the market. Because of this it is very hard for us to find talent in the market. Most of the developers in the market are used to the bigger players. For Jiffy.ai, if you search a resume because you are trying to find someone who has used Jiffy.ai, you won't be able to find it. So, when we onboard a new person, we want them to learn this new system, but it is a bit hard for them to pick up because there are no external learning materials on the Internet.
For training, they provide the foundation and advanced training. If you have other issues, they have a support portal, which shows a brief summary of the features. It's not very extensive, like Google Cloud Platform. Sometimes there are things that may not be available in the portal. While other products will also not have available the information in their portal, other people know it. So, you don't have in the community discussions about solutions to a problem that would not be available in the portal.
For how long have I used the solution?
We started this project last year in May.
What do I think about the stability of the solution?
In terms of the portal’s stability, the system is quite stable. We almost never have downtime, and if so, it is very minimal. However, in terms of bot stability, it depends on the server. The bot sometimes gets stuck, then you have to restart the bot, which is something for them to improve.
What do I think about the scalability of the solution?
I do not see any issues in terms of scalability. We can automate a process for a certain department and that process can be very similar to a process of another department. We might need to just change it a little, so we can use the existing solution that we have created. For example, if we create a reconciliation, then the same engine can be used for any big reconciliation tasks in other departments not related to finance. It could be done for engineering, operations, etc. It is very scalable in terms of reusing the existing solution.
How are customer service and technical support?
They are still quite a small player. Because of that, they can focus on the customer a lot more. If I am comparing them to a bigger player or other players that we have worked with in the past as well, they are a lot more responsive, passionate, and focused on us. They help the customer.
Which solution did I use previously and why did I switch?
We can create almost any type of solution in a very inexpensive way. In the past, we bought software to do certain processes. However, with Jiffy.ai, we can build the same software at a fraction of the cost. We no longer had to buy this other vendor's software anymore, which we licensed every year. With Jiffy.ai, we just have to pay the setup costs in the beginning and have them do it for us. We wouldn't have to pay them if we are doing it by ourselves. If you just use their service and do the setup ourselves, then we don't have to pay for the service, we would just need to pay for the service to use the Jiffy.ai platform to build our software. So, in this example, we are actually saving 97 percent of the costs.
How was the initial setup?
The deployment process is quite fast. Because they are small, they could focus on us. With the development, there are not a lot of processes to do it. We just have to set up the server. We can use their cloud, as cloud hosting or hosted in our on-premises. Even if it is hosted on-premises, the setup is quite fast. Training our staff was also quite fast. I didn't have any issues. I was quite happy with the setup.
The initial setup could take about a month or less, but we also had the incremental setup for our sister company. So, we have multiple entities in our company. The first time that we set up, we set up from scratch so there were a lot of other things that we needed to set up, but setting up another tenant for our sister company took a few days.
What was our ROI?
The reduction of work on a manually basis by project is between 50 to 90 percent. There are some processes where we almost automate the whole thing, and we just need manual handling by a person in certain rare situations. In that case, the reduction could be 90 to 99 percent. However, for certain processes, we can only automate 30 to 50 percent because the rest of the process still needs to be done by a human because of regulatory purposes, etc. So, it's a huge range: 30 percent to 99 percent.
What's my experience with pricing, setup cost, and licensing?
The pricing is quite competitive. As a small player in the market, they are quite aggressive in their pricing. With the features that they offer, it is quite worth the value.
Which other solutions did I evaluate?
Before we chose Jiffy.ai, we looked into other solutions, especially bigger, more established solution providers, like UiPath, Blue Prism, and Automation Anywhere. In terms of simplicity of usage, Jiffy.ai is easier to use since they are on a webpage. We put a portal on it and everything is available there. The UI is a bit more user-friendly and intuitive.
In terms of trying to do end-to-end process automation and how easy it is to do it, these are big pros and cons when compared to UiPath. In some ways, they are easier, and in some ways, they are not. I like with Jiffy.ai that we can use Python, but with UiPath, we can't use Python and need to use .NET. I'm unsure if they have enabled Python now. We also have a lot more flexibility with Jiffy.ai, e.g., we can connect to Google or any kind of system without having to do integration. We can just go from the front-end and record it. UiPath has this as well. You need to install Orchestrator on your PC. Then, you can install the design anywhere, because it is web-based, which is an advantage.
In other solutions, you have to install and set it up. If I have a new developer come in, then I have to install the system on their laptop before they are able to do their work. With Jiffy.ai, you can do it anywhere, on any laptop, as long as the laptop has access to the webpage. You just need access to the webpage, then you are able to do it. We control it from the portal as well. So, if I want to shut down or restart the bot, then I just have to go to the portal. I don't have to go to somewhere else, log into the server, or remote desktop to several laptops to do it. Everything is centralized on one laptop in one portal: the user access, the bot management, the task management, and the user interface for the human to manually handle certain stuff. Everything is on one page. This is an advantage over other solutions.
What other advice do I have?
You have to be open to trying new things. There are certain things that if you are already used to other bigger players in the market, then there are things that you like and things that you don't like. However, even the things that you don't like, it is mostly because you are already used to the way the service player is doing it. Therefore, if someone is doing it differently, it could be actually better, though it may not feel like it. I think you will find it exceeds your expectations.
Even with using humans, we have multiple redundancies to ensure there are no errors. The end results are not a lot of errors, though using the bot reduces the redundancy in having people check each other's work.
We are still reducing the full-time employees doing the work, but not up to 100 percent. We still need to maintain certain people for handling tasks that can't be handled by the bot, like manual exception and manual handling. Therefore, we cannot 100 percent automate everything. There are certain scenarios that require human judgment, preventing us from using the bot to do them.
I would rate this solution as an eight (out of 10).
Which version of this solution are you currently using?