Since last year, GenAI has been the topic #1 in the industry, everyone is talking about it, and first real implementations of this technology have started to pop up.
But adoption has not been easy, mainly because of two reasons: firstly, teams need to learn how to use GenAI; and secondly, they need to understand how to implement GenAI.
There are lots of moving pieces in an implementation of GenAI, first you have the input, normally in the form of prompt, but most companies will require this step to be fully automated, therefore the use of APIs will be required. Once GenAI does its job and comes back with an answer, that response will need to be processed, and that will need to be automated too.
Having a prompt where you can ask questions on demand is not enough for business purposes, a full automated process is required, where:
· Models can be trained with internal corporate data (and making sure that data does not leave the company)
· APIs are used to interact with the GenAI tool (input and output)
· Output from GenAI is automatically processed by other tools such as Slack/Teams, AIOps solutions or even ServiceNow
It’s not a secret that there are several companies that are delaying GenAI projects, even when they have budget allocated to this initiative, because they don’t have resources that know how to implement this kind of technology and, also, because don’t have enough time due to other priorities. This is being a big problem for those companies, because their competitors are already taking advantage of the productivity gains that GenAI comes with.
Democratize the access to GenAI
One way of fixing this problem is by leveraging no code automation platforms that will help you democratize the access to GenAI, meaning that everyone will be able to implement GenAI in their team.
The main problem that we are facing here is that teams that want to implement GenAI don’t have the expertise to do so, and teams that do know how to use GenAI do not have the SME knowledge on the specific topic where we want to implement GenAI.
For example, let’s review a real use case, IT outages, that are so common nowadays. These outages are having a big (negative) impact in companies, and every single company needs to reduce them or even completely eliminate them. However, finding the root cause of the outages is not an easy task.
Operations teams use a mix of approaches to proactively detect and reduce outages, including checking documentation, tribal knowledge (that person in the company that has seen everything and knows how to fix everything), reviewing what worked and what didn’t work in previous tickets… but this is a very long process, that every single company is trying to speed up.
IBM Rapid Network Automation
Using no code automation solutions like IBM Rapid Network Automation can help teams to deploy GenAI faster and effectively. Let’s see an example.
In this scenario, we are using a network performance monitor tool (IBM SevOne) to monitor the network, and SevOne just detected a problem with one of our firewalls, where the CPU used on a specific process (SSL process) is quite high.
Without GenAI, the normal process to investigate these issues would be raising an incident in ServiceNow (or any other ITSM solution), however the information available in the incident is quite limited, because the NPM tool only reports what is detected, that there is a problem on the CPU utilization of a process on a firewall.
ServiceNow Incident with GenAI disabled
With no further information about the possible reasons why we are having high CPU or how to fix it, it will take a long time for our operation teams to solve the problem, possibly too late, when this issue has become an outage.
This is a situation where GenAI can help us to enrich the incident data by providing us more context on potential reasons on why this is happening and potential solutions to this problem. GenAI could leverage thousands of articles, internal documentation, and successful resolutions that worked in the past to provide us more context on this incident.
But, as mentioned before, teams that deal with these issues usually don’t know how to use GenAI or even how to automate this process. Hence, using no code platforms that includes the APIs of the main vendors certified out of the box like IBM Rapid Network Automation, we can build in minutes a workflow that uses GenAI to enrich the incident, providing more and higher quality info to the teams.
Example of RNA Workflow Using GenAI and ServiceNow Integrations
This is an example of a real workflow created with IBM Rapid Network Automation that uses GenAI (in this case watsonx) to enrich the data available before raising the incident in ServiceNow.
In this workflow, the user of the tool wanted to be able to choose whether to enable or disable GenAI depending on the incident created, therefore the first operation the workflow does is to check if we want to use GenAI or not. If we don’t want to use GenAI, we will create the incident in ServiceNow straightaway. Basically, this is the original situation where we are not using GenAI to enrich the incident.
If GenAI is enabled, the next step will be sending the information from SevOne to the GenAI platform (watsonx), that will process the request using the model that we have chosen, and send a response that will be automatically sent to ServiceNow to create the incident.
Simplifying Working with APIs
Even creating the incident in ServiceNow will be very easy, the team won’t need to learn how the ServiceNow API works, or how to work with complex json/xml bodies. IBM Rapid Network Automation removes all that complexity by certifying APIs (more than 200 vendors are certified out of the box), meaning that every single action available in the API will be available as a building block with human readable content.
GenAI Outcome
Let’s compare the information available in this new incident using GenAI. The new enriched message is more valuable, as it clearly indicates the potential reason of the problem and how to fix it.
ServiceNow Incident with GenAI Enabled by RNA
Bear in mind that the workflow used for this incident can be used for any other type of incident. This is another example where there were issues with QoS.
QoS Incident with GenAI Disabled
The incident created not using GenAI does not bring too much information, but the incident with GenAI enabled clearly states what is happening and even includes a link with more information.
QoS Incident with GenAI Enabled
This is a real example of an automation workflow that leverage GenAI using the no code automation platform IBM Rapid Network Automation. A project that would have taken months to be completed (or not completed because of the complexity) without this capability, it took around 30 minutes to be built by a person that didn’t have any GenAI or scripting skills, saving thousands of man hours, and avoiding outages on the network, all thanks to no code automation with out of the box certified building blocks.
#TechnicalBlog