BeyondTrust standardizes on Certinia to connect resource utilization to customer satisfaction

Digital connection has brought significant change to how professional services are delivered in the technology industry, particularly in the wake of the pandemic. Being forced to deliver services remotely rather than on-site accelerated the adoption of new ways of working that were probably long overdue, but had not previously taken hold through force of habit. But after seeing the impact on resource utilization and customer experience, few companies are going back to their old ways. Sean Cashin, Senior Vice President, Customer Success and Services at cybersecurity company BeyondTrust, says:

We were able to really improve our availability and utilization numbers, because there was less time being spent on aeroplanes.

Now that an implementation doesn't have to be done in the course of a single engineer visit, the work can be spread over a period of time, which gives customers more space to reflect on all the information they have to absorb. He explains:

If I'm flying somebody [to the customer site], they're there with them all day, every day, for days, and it could be like drinking from a firehose. There's a lot of decisions, a lot of information to digest and everything.

Now, if we're able to do half days, instead of being with them for five full days, I'm with them for 10 half days. Or maybe it's five half days, a two week break, and another five half days.

The customer has longer exposure to professional services, greater chance to get comfortable with what it is that they're doing, what it is that they're trying to accomplish. [They can ask] more questions and [have] more quality time with the engineer, rather than it just being an info dump-and-digest.

That's important for the type of implementation BeyondTrust is delivering. With 20,000 customers worldwide, the company is a leading provider of Privileged Access Management, which adds extra protection around user accounts with high levels of access to digital assets, such as network administrators and other key roles. Typically privilege management is installed as part of a suite of products that also encompass identity management and vulnerability management. Many customers are in regulated industries where such precautions are mandatory, such as banking and other financial institutions, insurance, manufacturing, and so on.

While the implementation projects are not long — typically less than a day a week spread over at most a four- to five-month period — much of the work is consultative rather than simply setting up the technology. He says:

There's a certain amount of implementation, setting up whatever needs to be set up from a software perspective. But the majority of my engineers' time is spent helping the customer understand what they're being presented with, what decisions are the right decisions for them.

Security is an ongoing journey that never ends, but it's also a series of balancing choices between security and ability to function. It's very easy to completely lock down an environment so that nobody can do anything wrong. The challenge there is that it frequently makes it impossible to do anything right. You're just kneecapping the entire company's productivity. So our primary value as a professional services organization is helping customers understand those choices and what they need to do in order to continue to march down that journey of security maturity.

Shortly after Cashin joined the company in 2018, acquisitions and mergers brought four separate professional services organizations together under the BeyondTrust name. While two were already using the Professional Services Automation (PSA) solution from Certinia — then known as FinancialForce PSA — another was using a homegrown system and the fourth was still working off spreadsheets. Even the existing Certinia instances were not using it to its full potential, he adds. But the product's ability to provide a detailed view of schedules across a variety of different delivery channels, from video calls to on-site visits, was clearly what the organization was looking for. He goes on:

We have to know not only what a proserv engineer has to have done by what date. It becomes a matter of, we have to know, Tuesday morning from 10am till noon, Central Standard Time, engineer X and project manager Y are with customer Z. We had to have that level of granularity and accuracy and agility within our scheduling. And it had to be something we could easily view.

After standardizing on the Certinia PS Cloud product across the entire proserv organization, BeyondTrust saw a 20% increase in resource utilization. As Cashin points out, that's good news both for the company and for its customers:

To make maximal use of billable hours and utilization is every proserv organization's north star. If there's one KPI we're all going to have in common, it's billable utilization. We needed to be able to say, 'Oh, we just had a cancelation with this person. We can move them over to this project.' Or, 'Things aren't moving quite as quickly in the UK as they are in the East Coast of the US. We're going to start using the UK engineers, their afternoons, US mornings. That way, we're maximizing utilization and minimizing our customer's time to value — one and the same, really, when it comes down to it.

Adapting the Certinia application to BeyondTrust's needs was made easier because it's built on the Salesforce platform, which means that the company's developers already have the necessary skills to make adjustments. Cashin says:

I can take advantage of the internal Salesforce development team to continue to add fields here, do automations there, to be able to make sure that we were making maximal use of it. The way that I described it to the Certinia people is that I like how plastic it is. We can reformat it into whatever it is that we might need ... It gives me the opportunity to take advantage of internal talent and development to shape it into my unique interests.

This also helped speed the project to move everyone onto a unified system. This included adding appointment types to the scheduling system to differentiate between, say, a kick-off meeting, an architecture or implementation meeting. He goes on:

Because of the fact that it was within the Salesforce world, we really needed minimal services from them in order to stand this up. Within nine months after the acquisition, we were solidly within PSA, all of the other tools had been retired, the project managers were no longer siloed into the legacy company world, they were able to manage all the projects of all the different product types. And we were able to really see a pretty significant upswing in utilization in legacy organizations and to be honest, in some cases for the very first time, see what utilization levels look like.

The scheduling system also takes engineer skillsets into account when allocating resources to projects. This is particularly useful given the range of products in the BeyondTrust portfolio. He explains:

My proserv organization is around 120 people, right now. We have about five core products, but say, seven or eight that are out there, in varying degrees of support. Not all of our engineers can possibly know all of our products, they just wouldn't have the subject matter expertise I would demand that we have in front of our customers. So they remain specialized in, say, three or four of the products. And I don't expect the project managers to know all of the skills of all of the engineers.

The ability to go into the resource planner and say, 'I need an engineer in the Middle East who has an advanced knowledge of privileged remote access and Password Safe filter,' it gets rid of all of the chaff. It gives you a clean view of the only engineers that are going to be able to do the work that you need done. And it shows what the next 12-16 weeks of their time is going to look like, so you can find that common ground and get it booked pretty quickly.

A recent addition to the system has been integration to Calendly, which has made it much simpler to set up engineer appointments. Previously this was done via phone calls or email threads. Now the process is automated. Cashin says:

We're able just to send a link to the customer, the customer is presented with real-time information that's based on Certinia availability data in Calendly, so that they can pick a time with an appropriately skilled engineer that matches their needs for their timeline and their availability. It's a one and done. It creates the assignment in PSA, it creates the calendar entry in the engineer's Outlook and the project manager's Outlook. So it's been a huge effort saving for us.

Another integration automatically sends a Gainsight survey to the customer once a project successfully closes, and the customer satisfaction scores are written back to the project record in PS Cloud and Salesforce. Anything less than a nine or ten sentiment score creates a call to action for the regional manager to follow up with the customer and find out what the company could have done better or what was the reason for the dissatisfaction. The Calendly initiative is one example of a project that was prompted by that feedback. Cashin says:

Although I'm big on effort saving and everything else, really what prompted us to that was, we were getting consistent feedback on, it takes too much effort to land on a time with your engineers, can we make this simpler?

Greater flexibility in how services are delivered and supported goes hand-in-hand with greater complexity in managing resources. It's no surprise that BeyondTrust found the need for a reliable and sophisticated resource scheduling engine that could adapt to its needs. The integration with Calendly illustrates how end-to-end processes continue to be improved as a result of listening to customer feedback — all crucial ingredients in keeping pace with the ongoing evolution of professional services.

Generative AI replaces crypto as tech’s latest gold rush

But, as with the cryptocurrency gold rush, scammers aren’t the only ones rushing in to capitalise on the new generation of machine-learning software, the so-called “generative AI” software that can perform tasks ranging from writing to poems and drawing pictures, to writing software and filling out forms.

The world’s biggest software companies, including Google’s parent Alphabet, Microsoft, China’s Baidu, Salesforce, even Meta itself, have all swarmed into the space opened up by OpenAI’s ChatGPT, too.

And researchers and executives from those companies have been leaving, seemingly in droves, to found their own machine-learning startups, often addressing niche applications of generative AI.

In February, just 68 days after OpenAI launched ChatGPT, Alphabet’s two main AI companies, Google AI and DeepMind, announced a generative AI service of their own, Bard, that would offer many of the features of ChatGPT, but would be based on a machine-learning model Alphabet had been working on for years: the Language Model for Dialogue Applications, or LaMDA for short.

The next day, Microsoft got in on the action, announcing that its languishing search engine Bing would be powered by OpenAI’s machine-learning software, which Microsoft had been backing since 2021, topping up that investment in January with a further injection of $14 billion.

“AI will fundamentally change every software category, starting with the largest category of all – search,” said Satya Nadella, Microsoft chairman and chief executive at the time of the Bing announcement.

Microsoft chief executive Satya Nadella has been accelerating its investments in artificial intelligence as competition in the field of generative AI heats up. Bloomberg

And it didn’t take long for that prediction to start to come true.

In March, China’s own search engine giant Baidu announced that it, too, would be coming out with a ChatGPT rival known as “Weixin” (or “Ernie”), promising that the bot would not only be able to come up with answers to complicated maths problems, but would also show them a step-by-step explanation of how it came up with its answers.

But, as Nadella suggested, using generative AI to power search was just the start of the applications to which the technology would be put.

In early March, the customer relationship management (CRM) software giant Salesforce swarmed into the generative AI space, too, announcing it was working with OpenAI on “Einstein GPT”, a chatbot that could help Salesforce customers send personalised emails to their own customers, respond to customer queries, and help software developers write code.

“The world is experiencing one of the most profound technological shifts with the rise of real-time technologies and generative AI,” Salesforce chief executive Marc Benioff said at the launch.

Significantly, Einstein wouldn’t just use the machine-learning models created by OpenAI, which ingested vast swathes of public data from the internet to create so-called “large language models” capable of accurately guessing the words that make up answers to questions coming from ChatGPT users.

Salesforce founder and chief executive Marc Benioff  Oscar Colman

Einstein users would also be able to re-train the machine-learning model in real time on their own, private data, helping to ensure that content generated by Einstein was relevant and up-to-date, Salesforce said.

Meta’s own work on large language models is designed to help with such re-training of machine learning models, too.

In February, it announced a large language model known as “LLaMA” (Large Language Model Meta AI) which would come in a variety of sizes, so researchers could choose the size that was most cost-effective to re-train.

“Even with all the recent advancements in large language models, full research access to them remains limited because of the resources that are required to train and run such large models,” Meta said.

“Smaller models trained on more tokens — which are pieces of words — are easier to retrain and fine-tune for specific potential product use cases. We trained LLaMA 65B and LLaMA 33B on 1.4 trillion tokens. Our smallest model, LLaMA 7B, is trained on one trillion tokens.”

The high cost of training new machine-learning models is, it happens, the factor that most separates the ChatGPT boom from the cryptocurrency gold rush that preceded it.

Where startups could create their own cryptocurrency at little cost, giving rise to more than 23,000 different cryptocurrencies by some estimates, creating a new large language model requires access to large amounts of data, and large amounts of computer power to process that data, both of which can prove costly.

That’s created a barrier to entry for startups, and in the months since ChatGPT generated global interest in generative AI, startups have tended to be well-funded, and founded by researchers and entrepreneurs who already have a track record in the technology.

In May, two Google researchers who kick-started the generative AI boom with their 2017 paper “Attention is All You Need” were reported to have received $US8 million ($12.5 million) in a funding round led by Thrive Capital, a New York-based venture capital firm that also backed OpenAI.

Those two researchers, Ashish Vaswani and Niki Parma, had actually left Google in 2021 to start another machine-learning company, Adept AI, which itself has raised $US415 million to create machine-learning applications for enterprise customers.

Their latest startup, known as Essential AI, was still in stealth mode at the time of writing, but it doesn’t have the same enterprise focus as Adept AI, according to reports.

Those aren’t the only entrepreneurs who have left major AI companies to found startups.

Noam Shazeer, who was listed as second author on Google’s seminal 2017 research paper (behind Ashish Vaswani and immediately ahead of Niki Parma) has founded Character.ai, a startup which has raised around $US200 million to build a chatbot that mimics the personality of celebrities or people with specific occupations.

Meanwhile, Aidan Gomez, listed as the sixth author on Attention is All You Need, recently raised $US250 million for his startup, Cohere AI, which builds tools similar to Salesforce’s Einstein chatbot, that businesses can deploy to make their processes more efficient.

(Indeed, Salesforce is reported to be one of the backers of Cohere’s latest funding round, possibly out of its $US250 million Generative AI Fund which Salesforce announced at the time it revealed Einstein.)

So great is the investment in generative AI, the market research outfit IMARC Group now estimates the global AI market to be worth $US70.9 billion, and growing at a compound annual rate of 30 per cent.

That would make it worth $US350 billion by 2028, IMARC Group estimated.