With the current state of artificial intelligence (AI), everyone who may be in a related domain has heard of it or claims to know some of it, some actually have seen/ used its benefits but everyone agrees it’s going to be a game changer.
Game changing fully baked-in AI products for the industry, like chatGPT from openAI, are still some distance away but that doesn’t mean there isn’t enough capability available commercially to make a difference. At this very moment, every industry in the world that has some sort of technical and digital process, can benefit from the nuanced hand of an experienced AI operator.
These mentioned AI operators can look like a management consultant but that’s often just the shiny vinyl coating on the actual engine. I’ll talk about one such example from the AI realm where use of smart tools changed the BAU (business as usual) approach of a bank while savings weeks in processing latency.
I was transferred to middle east to work for a banking client in their corporate headquarters as a consultant for a popular SaaS software. The bank had purchased a pre-developed retail credit portfolio solution, or so they thought. The incorrect expectation was that the solution was going to be a plug and play module which couldn’t be farther from the truth. This mismatch added a strong emphasis on my role and transformed it overnight from the small scope of a developer to a large amalgamation of a consultant, analytics expert, salesman and guide.
While the role was full of unknowns and challenges, it also put me in touch with people from many different departments, who came to rely upon me for accurate advise and actual feasible solutions. The last one is an important distinction because it is the singular biggest difference in how enterprise technology business works vs how AI, or someone well versed with AI, designs processes and solutions from scratch to fit their way of understanding and doing business. For someone used to only enterprise technology business, this is unheard of, and something close to magical.
One of the persons I frequently interacted with, was the head of the network IT team. One of his responsibilities was to monitor and maintain network loads and capacity usage of in-house servers.
There was a combined requirement from the data team and the leadership group to have the network usage statistics displayed on an existing powerBI dashboard. Sounds simple, right? It is. But when you factor in the fact that this was a bank without an in-house analytics team, you begin to realise that the task becomes as difficult as probably making all Ubers as self-driving in India… which is another way to say, it was a mammoth task.
With my guidance, the network IT person contacted the server monitoring software team. Their software had the ability to export a daily and a weekly stats report without much hassle, which meant that the raw data became easily acquirable and now the ETL and data preparation was needed before it could be added to the powerBI report.
Again, simple, right? Not when we factor in the banking environment. While the data transformation wasn’t complex, it was however time consuming and the only recommendation that was available was to hire a resource. While that sounded to me like an absolutely wasteful way to go, the option itself was removed when we found out that the timeline to hire and onboard a new resource had jumped upto 4 months because of some legal issues.
Back to the problem. I had recently implored my hiring manager to get me approval for a Microsoft PowerApps license. I was planning to use it for another ongoing project and the request was approved.
This is when I activated the AI operator skillset from my armoury. The required data transformation and preparation didn’t seem complex to me and so I started to test this. I wrote multiple “office scripts” in MS Excel. These are a part of the Microsoft PowerApps license and are a much more powerful replacement of VBA, along with the added functionality of being integrated online with other M365 services such as Outlook email, MS Powerpoint and others.
While it was time consuming, the development did complete and all the expected data prep could now be done in 4 to 5 mins immediately proceeding the click of a button without any manual involvement.
To make the process actually usable, I again pulled out the AI operator and consulting skillset. This time I created a new ‘flow’ in Microsoft PowerAutomate. This flow was ‘on’ 24*7 and had the permissions to read my incoming mail. It’s starting trigger condition was an incoming email from the network IT head’s email address with an excel attachment and a specific keyword in the subject line.
As soon as those conditions would be met, the flow would initiate, pick up the attached raw network stats, initiate the data processing through the prepared office scripts, upload the data-table post processing completion to the powerBI dashboard’s input schema and end with sending a confirmation response email to the original sender.
Like clockwork. Every time. Without fail.
Now the management consultant, or an enterprise salesperson might dress this up as RPA (robotic process automation) use case, or an existing functionality within their software – but the fact of the matter is that the software and functionality existed before – but the solution didn’t exist before, only problem. The problem was eliminated and the solution was made possible by someone who is an experienced AI operator.
This example also highlights how AI doesn’t need to be an overhaul.
It doesn’t even need to be complicated mathematics that may only be comprehensible by double PhDs from Ivy League. It can be a simple increase in efficiency. It can be a simple decrease in manual workload.
What is truly magical is how small additions in efficiency and small reductions in manual workloads over sustained periods of time, start to look like the evolution of the whole unit into something ultra efficient and highly productive, possibly even unrecognisable.
I have significant experience in deploying AI and AI based tools to improve processes, increase efficiency and decrease workloads. I have utilised my skills for banks located in middle east (Dubai, Bahrain), south east asia (Singapore) and in africa (Johannesburg – South Africa) in the recent years.
Along with traditional use cases of application/ behaviour scorecards, segmentation, credit scoring and portfolio analysis, I have often deployed non-conventional but effective solutions which are often extremely simple in nature.
Pulkit Saraf 