Last updated on May 13th, 2024 at 07:34 pm
(caption: clockwise from top left, the Sphere in Las Vegas, a customer case study from TriMedX, British Telecom CIO Hena Jalil, Chief Communications Officer Paige Young with CEO Bill McDermott, opening reception at Bouchon, attendees leaving the opening keynote)
How ServiceNow’s Investments in Generative AI Are Transforming the Company And the Industry
AI in the enterprise is a serious, transformative business, on a scale we have rarely seen. The question is always the same, “OK, but how?” and I’d argue we saw more real wow-how on the convention floor and in breakout rooms at KNOWLEDGE, ServiceNow’s annual customer conference, now in its 17th year and held at the Venetian in Las Vegas, than I’ve ever seen at tech customer conference.
When ChatGPT launched, the business model was presumed to be consumer-focused. The consumer launch was extraordinarily successful, but it was clear that it was being treated as more of a toy, a tech curiosity, the newest shiny thing. I would not say, necessarily, that has changed in the consumer business at all. It shouldn’t be discounted how liberating and enabling this can be for individual users. People can be individually much more successful at creative and coding pursuits when supported by AI.
But the impact that AI is going to have on the enterprise is a whole other matter. It’s already clear what is going to be transformed (everything) and how (though ongoing exposure of data wedded to discovery turned into processes).
It’s the biggest shift with the biggest impact B2B enterprise software has ever seen. No question. For once, the hype is huge and deserved. 3.5x ROI, 93% planned adoption, 80% of CEOs believe it will fundamentally change their industries – yes, all the important stats. But the real glimmer of longevity, value and genius is the employee satisfaction numbers.
They love it.
Let me repeat that in case it wasn’t clear: employees are beginning to love their enterprise software platforms.
How to Fall in Love with Software
And here’s the primary reason why. If you have ever worked with enterprise software, any of it, and God help me I have spent a lot of time with CRM systems and ERP systems and SharePoint, it regularly makes grown humans cry. It is typically unwieldy, inflexible, slow, and frustrating.
In years—maybe in months—people will be astounded to learn we ever spent so much time and wasted energy using “dumb software.” Getting enterprise software to do what you needed was a dark art, usually involving a secret relationship with one of the dev techs so you could bribe them to get your customer’s feature request prioritized.
That or wait hours, months, and years for new features that were always late, and always clunky until the kinks were worked out, which, if we’re talking ERP, could be months or years. If ever. An astounding percentage of ERP rollouts just completely failed. How many of us have gnashed our teeth in frustration when a critical process you relied on that’s been running for days just suddenly… freezes?
Suddenly you are yelling at your laptop out of frustration. (we’ve all done it!) And there’s no known time to resolution. We’ve all watched someone lose it because a function failed taking them back three steps or the two new platforms you just acquired to make expense management better end up adding six hours to every business trip because they say they work together, but what they haven’t told you that x, y, and z need to happen first before you can unlock this critical feature.
You can’t get from screen A to screen X without getting to screen B, you can’t perform the function that you need to get to screen D and everything is deeply inflexible.
In most platforms one must follow exactly the path that the designers put into place—and let’s be candid, experience design in the enterprise has not really been a huge area of focus (although it was improving) and did not seem for a long time like this was going to be possible.
Too many platforms, not enough insight
Does this sound familiar? You may have up to 30 different systems in one environment, and much of your work involves switching platform to platform. Even if it’s just two or three that’s an awful lot of time moving back-and-forth between different systems to either perform the proper functions or to get the right data.
How often have you tried to select the data you want in a report, anticipating the dazzling insights you’re about to capture, when it appears analytics platform A treats this data scenario quite a bit differently than B and you need a little piece of connector code written, and that will take 11 months?
It’s hard to describe this pain if you’ve never used these types of platforms. And if you have, what we saw at ServiceNow this week might actually make you cry.
So many of the most tedious enterprise admin problems are literally about to disappear, if they haven’t already. Not everything, and not all at once, but ServiceNow is making possible, with breathtaking speed, a kind of liberation.
It is making software platforms into intelligent and highly flexible processes that can be nearly self-running.
Imagine if the UX and interface of your CRM platform, your expense platform, and the other 32 platforms your people have to switch between (at a loss of hours of productivity), suddenly dissolved, and they were all replaced by something called Now Assist. Now Assist helps you start your day *not* by saying hi, have 27 lost luggage report tickets that need to be resolved and you figure out that you have to access customer data, support data, platform data, and six different workflows to even get close.
No, it helps you start your day by resolving six of the tickets through the massive amount of data a generative AI has ingested on this particular issue, and then, it shows you eight different performance insights and suggests you look at deploying x to get the specific data output needed. And your time to produce actionable lost luggage reporting pulled from 14 different systems has gone from two weeks to ten minutes.
How? Generative AI exposes the critical data and functions in existing software. Combined with ServiceNow’s workflow streams, a set of mapped out new workflows is generated.. It then presents a sample workflow change and reporting schedule to finally get to an automation of your lost luggage report, which AI now assembles for you every morning while offering suggestions for optimization each time and what you might try next.
You don’t have to go from looking in SharePoint for an article for three hours anymore. Now Assist has already cataloged your entire SharePoint instance AND your CRM data and is presenting you with some recommendations on how to generate better-combined data.
Are you weeping with joy yet? Yeah, AI is a bit of hyperbole, but also, is not. Because it actually does these things, nearly invisibly. Need something? NowAssist allows you to prompt a specific request “Can I see the biggest return we got from our conferences last year mapped against employee involvement and travel/expense/sales reports?
Who should we select to go next year and what results should we expect is no longer a pipe dream of impossible complexity. Now Assist knows which data it needs to generate these insights, then reserves those players’ calendars and arranges their travel.
This is real today.
Workflow Automation Plus AI
The deployment of AI via Service Now’s workflow and process management is the combination that makes this work. To summarize: workflow automation, ServiceNow’s greatest strength, now combined with generative AI is truly revolutionary, with the potential to liberate virtually every user.
Imagine UX as we know it evaporating, replaced by a platform where the user centrally designs their own environment based NOT on what the system can do, but what the user’s very specific need and desired outcomes are.
Is it artificially intelligent? It’s the wrong question. The answer to that doesn’t really matter anymore. It has so much data, and we’ll talk about the model and tech/info architecture in a future article, it has all the data needed to find the answers, and it finds them. It can’t think per se. But if it’s been trained well on the right data, it has so much information, (every business plan ever written!) that whether it can actually think and apply logic is, at this stage, irrelevant.
It has enough data that it can write for you, instantly, given the right prompts and parameters, a business plan so good you’d swear it came out of a top consultancy. (There’s a lot to say about this because while at a very different level, than it was before, conversation is still happening about whether or not generative AI is any kind of intelligence and for a very long time I personally said no, it’s not, it’s not intelligent, it’s machine learning on an epic scale and that
It doesn’t matter anymore. It’s intelligence in the way that it can assess a document or an image and then with a prompt, transform those assets and the process around it. One of the examples that we saw on stage was tracking and recovering lost luggage and putting that data into an intelligent workflow. The demo demonstrated how generative AI can be used to quickly identify where a lost piece of luggage has most likely gone and almost immediately reroute it.
We are starting to solve real, annoying human problems in ways we just could not before.
When it comes to enterprise, it’s very easy to be cynical. Many of us have heard “xxyyzz is the game changer” since PKI in 1998 and in every other instance, it ends up being so much less than the hype and even something like blockchain, as useful as it is, ended up with some fairly fundamental operational restrictions (a hash? that … people compete to solve? every time? let’s not even talk about the constant need to actually finding a single transaction on a ledger) that made it impractical for many of the purposes it was originally envisioned.
This is the real deal. It’s as real as it gets. When I spoke with Element AI CTO and head of AI engineering Jeremy Barnes, formerly of Montreal of course, and asked him when generative AI would start to tap out in terms of processing powers and capacity and he looked at me and said “I don’t know. Yeah, sure it will happen at some point. But we are just at the very beginning of understanding its utility. We are not concerned about capacity. We have years of exploring how this works before coming anywhere close to running up against capacity for all but the very biggest models.”
This is a platform that knows what you’re asking it to do and then makes suggestions about how to do it better or what to do next. When you combine that ability to take action with ServiceNow’s workflows you’re starting to talk about not only a completely different kind of management software but a semi-autonomous, self-driven business management environment that holds ongoing transformation into its DNA and shows you what to do next, or how to do what you are doing better.
It’s not about how you’re going to get the right CRM data for a series of customers out of the system and then do something with it, whether it analyzes their data, looks at ways that they could be served better based on analysis in other words, an incremental improvement on the old way.
It is an exponential improvement achievable through the distillation of trillions of data points ingested during LLM training so that it knows how to improve and what to present next based on real prompting.
Adding ServiceNow to the equation puts business utility and usability around generative AI, in an extremely smart way. A serendipitous meeting between technology and platforms, AI needs processes to make it usable and productive is context, supporting process, and specific utility. LLMs produce data that is a moment in time, contextless, prompted only, with no memory of what happened before or after. Platforms like ServiceNow (if there are others; I haven’t seen these capabilities elsewhere yet) have process. They have learned how to structure their use of themselves to extract exactly the utility needed and then extend those lessons into their customers as partners (more on partners to come). That’s a vision exercise and incredibly, given so much accomplished, we are at the very beginning, barely out of the gate.
Enterprise Software’s Rockstar
ServiceNow has another enormous asset in Bill McDermott. He is the ideal leader for this technology and this moment in time. After years at SAP running the biggest ERP company in the world, Bill is recognized as the enterprise software king because of his presence, his energy, his passion for changing business. He’s seen it all. And he has found the dream.
After so long working within the confines of ERP, an undeniably powerful but also limiting and constricting platform for business, he is throwing off the bindings. It must feel like a kind of freedom to go from the powerful restrictions of ERP to an environment where your experience of using software can be redefined on the fly. When he talks about what ServiceNow can do, there’s one of two expressions on his face: awe and excitement about what the platform can do, or disbelief that everyone isn’t doing it already.
Bill is the best hype beast in the business, and finally he has found a technology that lives up to his ability to promote it. “Why are you not doing this already?” he asked repeatedly from the keynote stage in his trademark dark glasses.
After what I saw this week, I can’t imagine why you won’t. It’s turning the clunky inscrutability of data and process into “a single pane of glass” (the descriptive phrase of the conference) that will eventually expose the entirety of an enterprise’s data, and use it to reinvent it in ways we can’t possibly imagine today.
You could conclude from just an hour at ServiceNow’s that Generative AI is already changing B2B like no other technology has: not ERP, not blockchain, not even the internet or mobile. Why? These technologies augmented existing business practices and in places made them better, smarter, or widened access.
But generative AI operates within a business at a completely different level of strategic function. It contains so much knowledge and information, it is able, with staggering speed, to be strategic, to identify the optimal changes to how a company operates. and make it so much more efficient while freeing employees from the most tedious tasks. And yes, it is a form of intelligence. What form I can’t say. Do we have a word for it? I’m not sure. But the canyon between theirs generation and fourth is
No, AI cannot actively think per se and it is not a “creature”, but at the investment levels made, the amount of data it has been trained on, ingests and uses, combined with the sophistication of the algorithm, it can be hard to tell the difference, and the distinction may not even be meaningful.
I don’t have numbers yet on how much business got done at KNOWLEDGE, but the mood was very 1999, high optimism and excitement among employees, partners and customers. Interest was off the charts, and every employee I talked to was so excited to be there and so committed to the mission. And yes, it’s very easy to be skeptical about big new technology buzz, what else can they say, but after four days together it’s impossible for an entire conference to fake a mood.
Phrases from “once in a generation” to “the most important technology since the Industrial Revolution” were flying around and the thing that never happens happened: the KPIs and case studies lined up to the hype. The customers onstage had the glow of the recently promoted. Bill McDermott’s excitement was palpable, as evidenced by his clear disbelief that there were companies who wouldn’t do this.
As Canada’s country manager Chris Ellison said to me in an interview that will be up shortly “Canada has a productivity crisis and ServiceNow and technologies like it can play a role in changing that.” AI is a productivity acceleration agent and if you’ve seen our productivity numbers lately, you’ll now how critical this is to Canada right now.
The marriage between process automation and generative AI that was on display at KNOWLEDGE 24, is not complete, (disclosure: the company covered all my costs for this trip, but I would have paid to go to this one, and it would have been worth it), but it is a clear and highly compelling vision of the future of business.
By the Numbers:
- $275 billion: the size of ServiceNow’s TAM (total addressable market)
- 5000: number of innovations launched on the platform in seven months.
- Six: number of hours CSRs got back a week after implementing
- 5: percentage productivity improvement in a CSR environment
- 343: number of ServiceNow employees in 1994 20,000: members of the ServiceNow community at Knowledge24
In part two we’ll look at what all this actually translates to in practical terms with a look at some of the impressive customer stories and demos we saw at Knowledge24.
