The service portal is dead.
There, I said it.
Plain and simple – the employee self-service portal is an outdated technology that’s better left in the dumpster.
A few years ago, IT teams would dread being asked about email-based services support. They had to deal with 100s of emails asking basic, repeat questions.
“I got locked out of the system.” “My account is blocked.” “Please help me get a new laptop.” “I need access to project management software.”
Everything was in an email.
To tackle this, IT and Ops teams launched self-service portals to help employees with simple queries.
Some organizations spent millions – yes, millions of dollars to buy the “Snow” flake of software that takes years to implement, only to eventually realize that they made a huge mistake. Just because they had enough money to burn, organizations went with the market leader, thinking it would solve all their employee service problems.
Which, in due course, came back to bite them down the line.
Let me ask you something. Think about it honestly – as a customer, would you prefer to go to a company’s support portal, log in, and access your support tickets to see the status of your order, or would you prefer to work it out via a chatbot or even a simple messaging experience on their website or mobile app?
Which approach sounds better and much easier? I’m sure, you will agree that it is the latter.
And if your customers can have a better user experience, why should your employees be forced to bookmark or remember a URL to log in and check the status of their tickets?
Really, why the discrimination? Is it because of the capability or is it because of the “Snow” flake’s maturity (or lack thereof)?
The maturity model was invented to identify where you are, where you should go, and how to get there. Don’t overthink it.
Look at where you are today. Do you need this “Snow” flake software or can you use something better-suited for your organization?
Here is a simple data point for you.
The engagement metrics speak for themselves - Slack and Teams show so much engagement because it means real-time communication and faster resolution of queries, and it is all in one place.
The Microsoft TEI report shares some great metrics on the usage and effectiveness of the all-in-one solution. The stats are amazing in this report. Not only are employees able to get faster results, but organizations are also saving so much on $$.
I am not here to validate the value proposition of using a messaging channel for employee support, I am just here to say – serve your employees, where they are. 😊
Let's look at the ITIL V4 Model.
Pay attention to the second principle – Start where you are. Now, where are your employees? They are in Slack, MS Teams, or Google Chat. So, why not bring the helpdesk to those channels instead?
Introducing the conversational helpdesk.
Now, we all know, we’ve been scarred and burnt by chatbots that are boring, repetitive, and bring no little to no value to customers or employees.
They save a little bit of time for either party, by doing some basic chit-chat, so you don’t feel bad when you need support and no one is attending your call.
But ever since ChatGPT and LLMs were launched, the world of AI has turned upside down – isn't it? A few months ago, people hardly knew what Generative AI or Language models were, or that such a model even existed in the first place. But now, everyone talks about ChatGPT, right from children to CIOs.
Truly, everyone by now has heard and used ChatGPT, a truly conversational generative pre-trained transformer. That’s a Large Language Model that is built on 140Bn+ parameters, that’s like universal data.
And it keeps learning, with the next version being built on a trillion parameters.
When it comes to organizations, CIOs and CISOs aren’t comfortable with the use of ChatGPT, mainly due to the significant security concerns it poses. They have valid reasons to be cautious, as they want to ensure the protection of sensitive information such as employee data, customer data, and any intellectual property that belongs to the business.
The fear stems from the fact that open tools like ChatGPT have the potential to be leveraged for training the language model, thereby potentially exposing confidential information.
So what's the alternative?
What if there was a ChatGPT that learned everything about the organization – IT policies, HR policies, finance, travel, facilities… everything? What if there was a highly domain-specific model, that can become a true copilot for all employees?
What if there was a ChatGPT built specifically to suit your company, that can be used to onboard new hires, guide them on their 30-60-90 day plan, and answer important questions, ranging from their insurance to 401K policies, to IT-related questions about getting access to different tools and learning materials?
Now, some organizations have already tried building their own Slack bots or Teams bots. But that didn’t go well for them.
Everyone simply hated chatbots.
And the hatred is completely justified.
Because chatbots don’t have a basic understanding model. They are rule-based and are heavily focused on faster response. Their only model is to take questions, look/search existing rules, and give an answer - that’s it.
Enter GenAI. A perfect AI model, that not only provides faster responses but also has a teensy bit of empathy for user questions. That finds the closest of answers and if trained well on company data, does not need a rule, but a very precise answer.
Atom, our in-house GenAI is trained on the Internal Organization Model. Atom follows the most critical principle of ITIL 4 – “Start where you are”. Atom lives inside your MS Teams or Slack. Atom is omnipresent and always available.
Not only can Atom answer any question from the user, but Atom also understands multiple languages and responds back to the user in the same language to make sure they are more comfortable.
Atom learns from your organization's data models, whether it be on Microsoft Word, Notion, Confluence, Sharepoint, OneDrive, or anywhere.
Atom knows and provides answers to anything a user asks.
Can Atom also follow pre-defined processes?
Let's find out in my next blog post.