Hybrid AI in the Future of Work

 Hybrid AI in the Future of Work - ITOps Guest Post
This article was originally posted on ITOps Times

Due to ongoing improvements in artificial intelligence and machine learning technologies, we are on the cusp of an entirely new era in automation. Not only are software robots adept at performing routine, repetitive tasks on behalf of humans, but they are now capable of carrying out activities that rely on cognitive abilities, such as those requiring the use of judgment and emotion. One only needs to look at the cars we drive to recognize just how far automation technology has come.

Does this mean that there will be no place for humans in the future? The answer – at least for the foreseeable future – is a resounding no. That’s because, despite the growing list of benefits, there are also a number of drawbacks to having a system that is entirely autonomous. That’s where hybrid AI comes into play.

The concept behind hybrid AI is remarkably simple, even if the actual technologies and strategies driving it are incredibly complex. In basic terms, a hybrid model integrates humans throughout the automation process, but uses advanced technologies like deep learning and natural language processing to make automation systems even smarter.

AI needs humans
Beyond the hype, the truth is that artificial intelligence technology is simply not yet ready to replace humans – particularly when it comes to mission-critical applications. Take, for example, Tesla’s autopilot feature. While the vehicle itself is equipped with the capability to drive on its own, the driver behind the wheel is still required to remain alert and attentive to ensure his or her safety. In other words, AI is capable of running unassisted, but when it comes to mission-critical functions, it still needs humans, not only to train it, but to make sure everything stays on track.

The truth is, when artificial intelligence gets things right, everything is peachy. But when it doesn’t, the outcome can be disastrous – especially for larger organizations. And while modern AI may have some impressive cognitive capabilities, at the end of the day, it’s still just as its name indicates: artificial. Keeping humans in the mix ensures that the nuances of communication are present and that the output is accurate and relevant.

Humans need AI
On the other side of the coin, humans can benefit tremendously from artificial intelligence technology. And with 37% of organizations having already implemented AI to some degree, it’s clear that people and machines working side by side is becoming the norm rather than the exception. The reason being, artificial intelligence is like a force multiplier for human workers.

For example, data mining can be handled far faster and in much more massive volumes than any human being is capable of. Using AI, organizations can more effectively turn data into insights that can then be used to assist in human decision-making. This thereby drives innovation and competitive advantage.

Bringing it all together
As we progress toward a more automated future, a hybrid approach to integrating AI can help organizations figure out how to get from point A to point B with as little business disruption as possible. One way executives are handling the shift is to create automation centers of excellence (COE) that are dedicated to proliferating automation throughout the organization. Taking a structured approach like this helps to reduce confusion and limit friction.

Members of the COE are responsible for planning, ongoing testing and continuous oversight of the enterprise automation strategy. Typically, this group is made up of individuals who possess a mix of critical IT and business skills, such as developers, operations specialists and business analysts. Additionally, an entirely new role of automation engineer is being created to support the COE.

CIOs may choose to create their COEs with existing employees who are reskilled or newly hired team members. Regardless, COEs represent a strategic approach that is designed to drive adoption across the enterprise while delivering key results in support of company goals.

Ultimately, choosing a hybrid approach that includes a combination of humans and artificial intelligence, is simply the logical evolution of any disruptive technology. It safeguards against the risks of early-stage gaps and helps organizations get the most out of new solutions every step of the way. Done right, technology enables humans to focus on mission-critical applications while using AI to streamline operations and identify the best opportunities and strategies for ongoing organizational success.

AI is not an either/or proposition. It’s up to each organization to determine the right mix of humans and technology that makes sense. As new capabilities and options emerge, that mix will inevitably evolve. And the IT leaders that fully embrace their increasingly strategic value will know how to create the balance that will continually optimize and elevate staff, technology and the entire future of work.

This article was originally posted as a guest piece on ITOps Times. Click here to redirect to the official publication.

Want better self-service IT adoption? Try these 4 tips.

Many individuals (and even entire teams) mistakenly believe that self-service IT is something that threatens their livelihood. To the contrary, providing employees the control over their technology usage can make the job of IT much easier and more efficient. In other words, it’s a good thing, not something to fear and resist. So, how can a forward-thinking professional convince the powers-that-be that adopting intelligent automation is a step in the right direction?

Focus on the needs of the end-user.

The first part of the process involves identifying what needs end-users face that the IT department is responsible for fulfilling. This could include everything from simple password resets to entire user setups for new employees. As these needs are identified, they should be built out into what’s known as a self-service IT portfolio. The second part of the process involves determining the actions required in order to deliver these services. This will make up the service catalog.

Standardize and assign value.

With self-service automation, it’s important to ensure that any and all services and workflows being automated are as standardized as possible. Otherwise, you could end up automating broken processes, which will not only not help but could actually harm your overall business operations. It’s also important to assign a clear price/performance to each item in your service portfolio and catalog. This provides insight into the true value of the self-service IT activities.

Sell the benefits to each group.

If you want everyone – from the end-users to the IT team – to jump on the intelligent automation bandwagon, you have to demonstrate the actual benefits each group will achieve as a result. For instance, show employees how much more quickly they can get their needs taken care of without having to rely on someone from the help desk. At the same time, show IT personnel the time and effort they’ll be saving by eliminating these routine, repetitive tasks from their workload.

Start small and work from there.

You can’t expect a huge change such as self-service IT adoption to happen overnight. The process will take time and involve researching various automation platforms to determine which one best suits the particular needs of your business and then testing that tool before rolling out a full implementation. Start by automating one small area, such as password resets, and then work from there. Your service portfolio and catalog can provide the blueprint of what areas to automate in which order.

If you’re thinking of adopting intelligent automation to create a more consumer-style, self-service IT environment for your employees, it’s important to recognize that these things take time. Following the steps listed above can make the process go much more smoothly and help achieve the buy-in and support needed from others across the organization.

Ready to try intelligent, self-service automation? Click here to start your free trial.

eBook: 10 time consuming tasks you should automate

IT Incidents: From Alert to Remediation in 15 seconds [Webinar Recap]

Author: Guy Nadivi

Remediating IT incidents in just seconds after receiving an alert isn’t just a good performance goal to strive for. Rapid remediation might also be critical to reducing and even mitigating downtime. That’s important, because the cost of downtime to an enterprise can be scary. Even scarier though is what can happen to people’s jobs if they’re found to be responsible for failing to prevent the incidents that resulted in those downtimes.

So let’s talk a bit about how automation can help you avoid situations that imperil your organization, and possibly your career.

Mean Time to Resolution (MTTR) is a foundational KPI for just about every organization. If someone asked you “On average, how long does it take your organization to remediate IT Incidents after an alert?” what would your answer be from the choices below?

  • Less than 5 minutes
  • 5 – 15 minutes
  • As much as an hour
  • More than an hour

In an informal poll during a webinar, here’s how our audience responded:

More than half said that, on average, it takes them more than an hour to remediate IT incidents after an alert. That’s in line with research by MetricNet, a provider of benchmarks, performance metrics, scorecards and business data to Information Technology and Call Center Professionals.

Their global benchmarking database shows that the average incident MTTR is 8.40 business hours, but ranges widely, from a high of 33.67 hours to a low of 0.67 hours (shown below in the little tabular inset to the right). This wide variation is driven by several factors including ticket backlog, user population density, and the complexity of tickets handled.

Your mileage may vary, but obviously, it’s taking most organizations far longer than 15 seconds to remediate their incidents.

If that incident needing remediation involves a server outage, then the longer it takes to bring the server back up, the more it’s going to cost the organization.

Statista recently calculated the cost of enterprise server downtime, and what they found makes the phrase “time is money” seem like an understatement. According to Statista’s research, 60% of organizations worldwide reported that the average cost PER HOUR of enterprise server downtime was anywhere from $301,000 to $2 million!

With server downtime being so expensive, Gartner has some interesting data points to share on that issue (ID G00377088 – April 9, 2019).

First off, they report receiving over 650 client inquires between 2017 and 2019 on this topic, and we’re still not done with 2019. So clearly this is a topic that’s top-of-mind with C-suite executives.

Secondly, they state that through 2021, just 2 years from now, 65% of Infrastructure and Operations leaders will underinvest in their availability and recovery needs because they use estimated cost-of-downtime metrics.

As it turns out, Ayehu can help you get a more accurate estimate of your downtime costs so they’re not underestimated.

In our eBook titled “How to Measure IT Process Automation ROI”, there’s a specific formula for calculating the cost of downtime. The eBook is free to download on our website, and also includes access to all of our ROI formulas, which are fairly straightforward to calculate.

Let’s look at another data point about outages, this one from the Uptime Institute’s 2019 Annual Data Center Survey Results. They report that “Outages continue to cause significant problems for operators. Just over a third (34%) of all respondents had an outage or severe IT service degradation in the past year, while half (50%) had an outage or severe IT service degradation in the past three years.”

So if you were thinking painful outages only happen at your organization, think again. They’re happening everywhere. And as the research from Statista emphasized, when outages hit, it’s usually very expensive.

The Uptime Institute has an even more alarming statistic they’ve published.

They’ve found that more than 70% of all data center outages are caused by human error and not by a fault in the infrastructure design!

Let’s pause for a moment to ponder that. In 70% of cases, all it took to bring today’s most powerful high-tech to its knees was a person making an honest mistake.

That’s actually not too surprising though, is it? All of us have mistyped a keyboard stroke here or made an erroneous mouse click there. How many times has it happened that someone absent-mindedly pressed “Reply All” to an email meant for one person, then realized with horror that their message just went out to the entire organization?

So mistakes happen to everyone, and that includes data center operators. And unfortunately, when they make a mistake that leads to an outage, the consequences can be catastrophic.

One well-known example of an honest human mistake that led to a spectacular outage occurred back in late February of 2017. Someone on Amazon’s S3 team input a command incorrectly that led to the entire Amazon Simple Storage Service being taken down, which impacted 150,000 organizations and led to many millions of dollars in losses.

If infrastructure design usually isn’t the issue, and 70% of the time outages are a direct result of human error, then logic suggests that the key would be to eliminate the potential for human error. And just to emphasize the nuance of this point, we’re NOT advocating eliminating humans, but eliminating the potential for human error while keeping humans very much involved. How do we do that?

Well, you won’t be too surprised to learn we do it through automation.

Let’s start by taking a look at the typical infrastructure and operations troubleshooting process.

This process should look pretty familiar to you.

In general, many organizations (including large ones) do most of these phases manually. The problem with that is that it makes every phase of this process vulnerable to human error.

There’s a better way, however. It involves automating much of this process, which can reduce the time it takes to remediate an IT incident down to seconds. And automation isn’t just faster, it also eliminates the potential for human error, which should radically reduce the likelihood that your environment will experience an outage due to human error.

Here’s how that would work. It involves using the Ayehu platform as an integration hub in your environment. Ayehu would then connect to every system that needs to be interacted with when remediating an incident.

For example, if your environment has a monitoring system like SolarWinds, Big Panda, or Microsoft System Center, that’s where an incident will be detected first. The monitoring system (now integrated with Ayehu) will generate an alert which Ayehu will instantaneously intercept. (BTW – if there’s a monitoring system or any kind of platform in your environment that we don’t have an off-the-shelf integration for, it’s usually still pretty easy to connect to it via a REST API call.)

Ayehu will then parse that alert to determine what the underlying incident is, and launch an automated workflow to remediate it.

As a first step in our workflow we’re going to automatically create a ticket in ServiceNow, BMC Remedy, JIRA, or any ITSM platform you prefer. Here again is where automation really shines over taking the manual approach, because letting the workflow handle the documentation will ensure that it gets done in a timely manner (in fact, in real-time) and that it gets done thoroughly. This brings relief to service desk staff who often don’t have the time or the patience to document every aspect of a resolution properly because they’re under such a heavy workload.

The next step, and actually this can be at any step within that workflow, is pausing its execution to notify and seek human approval for continuation. To illustrate why you might do this, let’s say that a workflow got triggered because SolarWinds generated an alert that a server dropped below 10% free disk space. The workflow could then go and delete a bunch of temp files, it could compress a bunch of log files and move them somewhere else, and do all sorts of other things to free up space. Before it does any of that though, the workflow can be configured to require human approval for any of those steps.

The human can either grant or deny approval so the workflow can continue on, and that decision can be delivered via laptop, smartphone, email, instant messenger, or even regular telephone. However, please note that this notification/approval phase is entirely optional. You can also choose to put the workflow on autopilot and proceed without any human intervention. It’s all up to you, and either option is easy to implement.

Then the workflow can begin remediating the incident which triggered the alert.

As the remediation is taking place, Ayehu can update the service desk ticket in real-time by documenting every step of the incident remediation process.

Once the incident remediation is completed, Ayehu can automatically close the ticket.

Finally, Ayehu can go back into the monitoring system and automatically dismiss the alert that triggered the entire process.

This, by the way, illustrates why we think of Ayehu as a virtual operator which we sometimes refer to as “Level 0 Tech Support”. A lot of incidents can be resolved automatically by Ayehu without any human intervention, and thus without the need for attention from a Level 1 technician.

This then is how you can go from alert to remediation in 15 seconds, while simultaneously eliminating the potential for human error that can lead to outages in your environment.

Gartner concurs with this approach.

In a recently refreshed paper they published (ID G00336149 – April 11, 2019) one of their Vice-Presidents wrote that “The intricacy of access layer network decisions and the aggravation of end-user downtime are more than IT organizations can handle. Infrastructure and operations leaders must implement automation and artificial intelligence solutions to reduce mundane tasks and lost productivity.”

No ambiguity there.

Gartner’s advice is a good opportunity for me to segue into one last topic – artificial intelligence.

The Ayehu platform has AI built-in, and it’s one of the reasons you’ll be able to not only quickly remediate your IT incidents, but also quickly build the workflows that will do that remediation.

Ayehu is partnered with SRI International (SRI), formerly known as the Stanford Research Institute. In case you’re not familiar with them, SRI does high-level research for government agencies, commercial organizations, and private foundations. They also license their technologies, form strategic partnerships (like the one they have with us) and creates spin-off companies. They’ve received more than 4,000 patents and patent applications worldwide to date. SRI is our design partner, and they’ve designed the algorithms and other elements of our AI/ML functionality. What they’ve done so far is pretty cool, but what we’re working on going forward is what’s really exciting.

One of the ways Ayehu implements AI is through VSAs, which is shorthand for “Virtual Support Agents”.

VSA’s differ from chatbots in that they’re not only conversational, but more importantly they’re also actionable. That makes them the next logical step or evolution up from a chatbot. However, in order for a VSA to execute actionable tasks and be functionally useful, it has to be plugged in to an enterprise grade automation platform that can carry out a user’s request intelligently.

We deliver a lot of our VSA functionality through Slack, and we also have integrations with Alexa and IBM Watson. We’re also incorporating an MS-Teams interface, and looking into others as well.

How is this relevant to remediating incidents?

Well, if a service desk can offload a larger portion of its tickets to VSA’s, and provide its users with more of a self-service modality, then that frees up the service desk staff to automate more of the kinds of data center tasks that are tedious, repetitive, and prone to human error. And as I’ve previously stated, eliminating the potential for human error is key to reducing the likelihood of outages.

Speaking of tickets, another informal webinar poll we conducted asked:

On average, how many support tickets per month does your IT organization deal with?

  • Less than 100
  • 101 – 250
  • 251 – 1,000
  • More than 1,000

Here’s how our audience responded:

Nearly 90% receive 251 or more tickets per month. Over half get more than 1,000!

For comparison, the Zendesk Benchmark reports that among their customers, the average is 777 tickets per month.

Given the volume of tickets received per month, the current average duration it takes to remediate an incident, and most importantly the onerous cost of downtime, automation can go a long way towards helping service desks maximize their efficiency by being a force multiplier for existing staff.

Q:          What types of notifications can the VSA send at the time of incident?

A:           Notifications can be delivered either as text or speech.

Q:          How does the Ayehu tool differ from other leading RPA tools available on the market?

A:           RPA tools are typically doing screen automation with an agent. Ayehu’s automation is an agentless platform that primarily interfaces with backend APIs.

Q:          Do we have to do API programming or other scripting as a part of implementation?

A:           No. Ayehu’s out-of-the-box integrations typically only require a few configuration parameters.

Q:          Do we have an option to create custom activities? If so, which programing language should be used?

A:           In our roadmap, we will be offering the ability to create custom activity content out-of-the-box.

Q:          Do out-of-the-box workflows work on all types of operating systems?

A:           Yes. You just define the type of operating system within the workflow.

Q:          How does Ayehu connect and authenticate with various endpoint devices (e.g. Windows, UNIX, network devices, etc.)? Is it password-less, connection through a password vault, etc?

A:           That depends on what type of authentication is required internally by the organization. Ayehu integrated with the CyberArk password vault can be leveraged when privileged account credentials are involved. Any type of user credential information that is manually input into a workflow or device is encrypted within Ayehu’s database. Also, certificates on SSH commands, Windows authentication, and localized authentication are all accessible out-of-the-box. Please contact us for questions about security scenarios specific to your environment.

Q:          What are all the possible modes that VSAs can interact with End Users?

A:           Text, Text-to-Speech, and Buttons.

Q:          Can we create role-based access for Ayehu?

A:           Yes. That’s a standard function which can also be controlled by and synchronized with Active Directory groups out-of-the-box.

Q:          Apart from incident tickets, does Aheyu operate on request tickets (e.g. on-demand access management, software requests from end-users, etc.)?

A:           Yes. The integration packs we offer for ServiceNow, JIRA, BMC Remedy, etc. all provide this capability for both standard and custom forms.

Q:          Does Ayehu provide APIs for an integration that’s not available out of the box?

A:           Yes. There are two options. You can either forward an event to Ayehu using our webservice which is based on a RESTful API, or from within the workflow you can send messages outbound that are either scheduled or event-driven. This allows you to do things such as make a database call, set an SNMP trap, handling SYSLOG messages, etc.

Q:          Does Ayehu provide any learning portal for developers to learn how to use the tool?

A:           Yes. The Ayehu Automation Academy is an online Learning Management System we just launched recently. It includes exams that provide you an opportunity to bolster your professional credentials by earning a certification. If you’re looking to advance your organization’s move to an automated future, as well as your career prospects, be sure to check out the Academy.

Q:          Does Ayehu identify issues like a monitoring tool does?

A:           Ayehu is not a monitoring tool like Solarwinds, Big Panda, etc. Once Ayehu receives an alert from one of those monitoring systems, it can trigger a workflow that remediates the underlying incident which generated that alert.

Q:          We have 7 different monitoring systems in our environment. Can Ayehu accept alerts from all of them simultaneously?

A:           Yes. Ayehu’s integrations are independent of one another, and it can also accept alerts from webservices. We have numerous deployments where thousands of alerts are received from a variety of sources and Ayehu can scale up to handle them all.

Q:          What does the AI in Ayehu do?

A:           There are different areas where AI is used. From use in understanding intent through chatbots to workflow design recommendations, and also suggesting workflows to remediate events through the Ayehu Brain service. Please contact an account executive to learn more.

New call-to-action

5 Mistakes to Avoid with Self-Service Automation

Self-service automation is becoming more of the norm rather than the exception. In fact, a recent survey by SDI found that 61% of businesses were focusing on some type of self-service initiative (up from 47% in 2015). And it’s not only for making your customers’ lives easier. Many organizations are realizing the benefits of providing self-service options to employees to eliminate the need for many of the common issues plaguing the help desk, such as password resets and system refreshes. If you’re thinking about jumping on the bandwagon, here are a few common mistakes you should actively avoid.

Inadequate Communication – If you want your employees to adopt and embrace self-service technology, you have to ensure that they understand its many benefits. This is particularly important for your IT team, some of whom may feel uneasy or even threatened by the thought of automated technology handling some of their tasks. Gain acceptance and buy-in by communicating how self-service options will actually make the lives and jobs of everyone easier and more efficient.

Lack of Knowledge – What types of activities can you – and more importantly – should you be transitioning over to self-service? Many otherwise savvy IT decision makers rush into self-service implementation before they truly have a good understanding of what tasks are most beneficial to automate. Take time to learn about what your IT team is bogged down by and also what areas the end-user might not only benefit from, but actually appreciate the ability to handle things on their own.

Not Choosing a Tool Carefully – Not all self-service automation platforms are created equal and if you don’t carefully and thoroughly do your homework, you could end up with a less-than-ideal result. Not only does implementing a faulty tool mean more headaches for your IT department, but the frustration of everyone who has to use it will ultimately lead to disengagement, resistance and/or complete lack of adoption. Make sure the platform you choose is robust, user-friendly and versatile enough to handle both full and semi-automation needs.

Setting and Forgetting It – Like anything else in technology, self-service automation isn’t something that you can simply put in place and never think about again. Not only is it important to keep up to date from a tech standpoint, but it’s equally important to ensure that the system you have in place remains as effective as possible. Conducting regular audits of both the IT department and the end-users can help you determine whether new tasks could be automated or if existing ones could use some tweaking.

Forgetting the Intangibles – Last but not least, maintaining an environment in which self-service automation is embraced and celebrated involves regular assessment and selling of the many benefits this technology provides. When calculating ROI, don’t forget to also consider the intangible ways self-service is good for your organization, particularly how it allows IT to improve its meaningful contribution to the organization. That is a value that can and should be recognized across the board.

What could self-service automation do for your company? Why not find out today by starting your free 30 day trial of Ayehu. No obligation, just enhanced efficiency and better overall operations. Get your free trial now by clicking here!

Creating an Intelligent Virtual Support Agent for your ITSM

Author: Guy Nadivi, Sr. Director of Marketing, Ayehu

A topic that’s really getting a lot of buzz these days is Virtual Support Agents.  Virtual Support Agents – or VSAs – are the next logical evolution of a chatbot, because where chatbots are primarily conversational, a VSA is both conversational and actionable, making them much more valuable to an enterprise, and particularly to a service desk using an ITSM platform.

In order to better understand why VSAs are so top-of-mind, it can be helpful to take a step back and understand what’s happening right now in IT operations at enterprises around the world, particularly with all the data they’re dealing with.

Do you know what a Zettabyte is? NO GOOGLING!

A Zettabyte is a trillion Gigabytes. That’s a “1” followed by 21 zeroes.

As humans, it can be hard for us to wrap our minds around numbers that large, so let’s use a visual metaphor to help provide a frame of reference for how much data 1 Zettabyte represents.

The visual metaphor we’d like you to envision should be a familiar one – grains of rice. For the sake of this visual comparison, let’s say one grain of rice is equal to one byte of data.

That makes a kilobyte, a thousand grains, equal to about a CUP of rice.

A megabyte of data would then be the equivalent of 8 BAGS of rice.

1 gigabyte of data in terms of rice is equal to 3 container TRUCKS.

A terabyte of data would then be the equivalent of 2 CONTAINER SHIPS full of nothing but rice.

Now get this – an exabyte of data in terms of rice, would cover all of MANHATTAN.

A petabyte of data, would just about cover all of TURKEY. (BTW – turkey & rice is a great combo!)

Finally, here’s what a Zettabyte of data in terms of rice would do.  It would fill up the PACIFIC OCEAN! 

Now, this last visual using the Pacific Ocean is very relevant, especially if you work in IT operations.  That’s because you literally feel like an entire ocean’s worth of data is inundating you these days, thanks to all the systems you’re maintaining that create, store, access and deliver data for your employees, customers, partners, etc.  Life in IT operations is a relentless tsunami of incidents, events, thresholds being crossed, etc., and it’s only getting worse.

How much worse?

In 2017 The Economist published a chart produced by IDC in conjunction with Bloomberg estimating the size of data comprising “The Digital Universe”. In 2013 there were 4.4 Zettabytes of data worldwide, but by 2020 there will be 44 Zettabytes. That’s an astounding CAGR of 47%. Don’t expect things to slow down though, because it’s estimated that by 2025 there will be 175 zettabytes of data worldwide!

Interestingly, IDC/Bloomberg discovered what appears to be a correlation between the exponential growth of data and an increase in the number of times companies are mentioning “artificial intelligence” in their earnings calls. This is probably not a coincidence, and it underscores something we say over and over at Ayehu – people don’t scale very well. 

Even the very best data center workers can only do so much. At some point, and that point is pretty much right now, end user self-service in concert with automation has got to take on a greater share of the service desk tasks all this growth in data is necessitating.

So when does it make the most sense for your end users to interface with Virtual Support Agents instead of live operators at the service desk? Well, one obvious answer to that might be as your organization’s first point of contact for L1 support issues.

We’re talking things like password resets, which Gartner estimates are responsible for as much as 40% of your service desk’s call volume.

Onboarding employees, and its closely related counterpart task offboarding employees, are also excellent tasks for a VSA.

How about VM provisioning and VM resizing?  These tasks lend themselves very well to a VSA interface.

Another good one is service restarts.  It makes a lot of sense to empower end-users to be able to do this themselves via a VSA.

And there are many, many more L1 support issues that would be great candidates for Virtual Support Agents.

But then what about your service desk?  What should your human agents do once the Virtual Support Agent has taken on all these tasks? 


Well, there are still probably a number of L1 support issues that are not well-suited for a VSA (at least not yet). The service desk can continue handling these.

Of course most, if not all of your L2 and L3 support issues are still best handled by a human service desk, for now.

And finally, vendor support issues are also probably still best managed by the service desk.

Still, deploying a Virtual Support Agent will clearly shift a lot of tedious, laborious tasks off of the service desk and free them up to do other things. But what kind of impact does this have on the cost structure of a service desk?

The average industry cost of handling an L1 support ticket is $22. In comparison, the average cost of a ticket handled by a Virtual Support Agent is just $2. Using these figures to do a back-of-the-envelope calculation of how many tickets can be off-loaded from the service desk to the VSA, will likely yield a significant reduction in support costs. What’s more, issues handled by VSAs often gets remediated much faster, resulting in greater end-user satisfaction than going through the service desk.

And if end users are going through a VSA instead of the service desk, then that means the VSA can reroute a trainload of L1 incident tickets away from the service desk, freeing up that staff to focus on more important, and more strategic things.  This is a huge value proposition!

What about the benefits VSAs provide to end-users?

Here’s an obvious one. Ask any end-user how much they love waiting on hold when they need an issue resolved, but no one’s available to help them. Waiting on hold can really degrade the user’s experience and perception of the service desk. With Virtual Support Agents, on the other hand, no one ever gets put on hold because the VSA is available 24/7/365. VSAs never take a break, or call in sick, or get temperamental due to mood swings. They’re always available and ready to perform on-demand.

Then there’s the mean-time-to-resolution metric, better known as MTTR. As every service desk knows, the speediness of their incident remediation outcomes is one of the major KPI’s they’re judged on.  Well when it comes to speedy MTTR, a VSA should be faster than a human being just about every single time.

Finally, does your enterprise have younger employees, particularly from the millennial generation? Of course it does! Just about every organization does, and some have them in far greater percentages than others. Well, guess what? This generation, raised on Facebook and mobile apps, generally prefers interfacing with technology as opposed to people. And that’s something we all know empirically because we’ve seen it for ourselves.

In addition to that visual proof, a survey conducted in 2018 by Acquire.io found that 40% of millennials said that they chat with chatbots on a daily basis! So, providing VSAs that empower younger workers with self-service capabilities might just give your organization a competitive advantage in attracting the best and brightest young talent to your company.

Questions & Answers

Q:          What are the pros and cons of using general purpose bot engines compared to your solution?

A:           General purpose bot engines won’t actually perform the actions on your infrastructure, devices, monitoring tools, business applications, etc. All they could really do is ingest a request. By contrast, Ayehu could not only ingest a request, but actually execute the necessary actions needed to fulfill that request. This adds a virtual operator to your environment that’s available 24x7x365. Additionally, Ayehu is a vendor-agnostic tool that is capable of interfacing with MS-Teams, Skype, etc. to provide a general purpose chat tool with intelligent automation capabilities.

Q:          Do you have on-premise solution?

A:           Yes.  Ayehu can be installed on-premise, on a public or private cloud, or in a hybrid combination of all three.

Q:          Do you have voice integration?

A:           Ayehu integrates with Amazon Alexa, and now also offers Angie™, a voice-enabled Intelligent Virtual Support Agent specifically designed for IT Service Desks.

Q:          If a user selects a wrong choice (clicks the wrong button) how does he or she fix it?

A:           It depends on how the workflow is designed. Breakpoints can be inserted in the workflow to ask the endpoint user to confirm their selection, or go back to reselect.  Ayehu also offers error-handling mechanisms within the workflow itself.

Q:          Does Ayehu provide orchestration capabilities or do you rely on a 3rd party orchestration tool?

A:           Ayehu IS an enterprise-grade orchestration tool, offering over 500 pre-built, platform-specific activities that allow you to orchestrate multi-platform workflows from a single pane of glass.

Q:          Can you please share the slide on IVA vs. ServiceDesk and elaborate a bit on the use cases?

A:           The entire PowerPoint file presented in this webinar can be found on SlideShare.

Q:          Can you explain in a bit more detail on intent-based interactions?

A:           Intent is just that: what the user’s intent is when interacting with the Virtual Support Agent (VSA). For example, if a user types “change my password”, the intent could be categorized as “password reset”. That would then automatically trigger the “password reset” workflow.

Q:          Can we use machine learning from an external source, train our model, and let Ayehu query our external source for additional information?

A:           Yes. Ayehu can integrate with any external source or application, especially when it has an API for us to interface with.

Q:          Can I create new automations to my in-house applications?

A:           Yes. Ayehu can integrate with any application bi-directionally. Once integrated with your in-house applications, Ayehu can execute automated actions upon them.

Q:          Is there an auto form-filling feature that can fill in a form in an existing web application?

A:           Yes. Ayehu provides a self-service capability that will enable this.

Q:          How can I improve or check how my workflows are working and helping my employees to resolve their issues?

A:           Ayehu provides an audit trail and reporting that provides visibility into workflow performance. Additionally, reports are available on time saved, ROI, MTTR, etc. that can quantify the benefits of those workflows.

Q:          What happens when your VSA cannot help the end-user?

A:           The workflow behind the VSA can be configured to escalate to a live support agent.

Q:          If there is a long list of choices, what options do you have? Dropdown?

A:           In addition to the buttons, dropdowns will be provided soon in Slack as well.

Q:          Did I understand correctly, an admin will need to create the questions and button responses? If so, is this a scripted Virtual Agent to manage routine questions?

A:           Ayehu is scriptless and codeless. The workflow behind the VSA is configured to mimic the actions of a live support agent, which requires you to pre-configure the questions and expected answers in a deterministic manner.

Q:          is NLP/NLU dependent on an IBM Watson to understand intent?

A:           Yes, and soon Ayehu will be providing its own NLP/NLU services.

Q:          Are you using machine learning for creating the conversations? Or do we have to use intents and entities along with the dialogs?

A:           Yes, you currently have to use intents and entities, but our road map includes using machine learning to provide suggestions that will improve the dialogs.

Q:          What are the other platforms from which I can deploy the VSA, apart from Slack?

A:           Microsoft Teams, Amazon Alexa, ServiceNow, ConnectNow, LogMeIn, and any other chatbot using APIs.

Missed the live Webinar? Watch it on-demand and see the above in action by clicking here.

New call-to-action