“People, Processes, Products”: Reflecting on Thoughts from 2016 and 2017 to Perhaps Anticipate the State of AI Adoption in 2020
I’ve become a history buff over time — reluctantly at first (middle school), tepidly soon after (high school), inexorably in college, and now, undeniably spending way too much nerding out about it. We often hear that history explains where we come from; why we act in the ways we do; and, sometimes, it can even outline what the future has in store for us (“history doesn’t repeat itself but it sure rhymes,” said our dear friend Mark Twain). History — the study and interpretation of past events — can stitch together linearity for cultures, governments, politics, and of course, business. In the spirit of reflecting on my recent professional past — personally looking back on my experiences at Xamarin, Microsoft, and now Algolia — I can’t help to think that we’re living in an ambitious yet ambiguous time (oversimplification certainly!). In retrospect, cloud computing has given rise to new emerging technologies once only figments of our imagination — think AI, ML, data analytics in real time, etc. At the same time, enterprise customers more than ever in 2019 see a paramount need to evolve their businesses, optimize their operations, and for some, come up with new commercial models in light of slowing productivity gains and increased competition. Their futures are not guaranteed by any means. While most understand they need to be become technology driven in order to positively impact their business lines, they are struggling to decide which initiatives to prioritize. How can we blame them, when “innovation” for them is touted in the same breath as buzzwords “Data”, “AI”, “Blockchain”, or my favorite, “computing on the edge?” — concepts they are perhaps unfamiliar with? As we will see, a certain BCG report[1] touts there is no denying that AI is a top of mind issue — but what does that entail? With these questions in mind, and in spirit of looking at the past to potentially guestimate the future, I’d like to take the occasion to reflect on some my thoughts put forth in in 2016[2] and 2017[3] (some that were off the mark), to reflect on why state of AI adoption in the enterprise is where it is in 2019, and hypothesize where it will go in 2020.
Mobile, Cloud, and Data oh my!
Over the last two years I made a series of arguments around the impact of mobile and cloud computing for the enterprise and how their partners should evolve their business models to address those concerns. I first argued that mobile would be a game changer for the enterprise as it offered new methods for brands to connect with customers and for employers to enhance employee productivity. In tandem I implied it would be one of the most profitable service centers for consulting companies, pulled by their customers’ need build to build mobile-cloud first experiences. My analysis was rooted in my perspective that enterprise customers had a deep bench of applications walled within their firewall perimeters in desperate need of modernization. That being said, I failed to understand that web apps’ infrastructure slowed down that migration because they took precedent over moving from a web form factor to a mobile one. Web interfaces were “good enough” in light of the gargantuan and more pressing challenge of modernizing the back end, even though ROI involved modernizing both at the same time was higher. My mistake: I had argued that the IaaS opportunity couldn’t materialize without a modernization of UIs. Clearly it could. In addition, I thought it in the best interest of consulting firms to promote both mobile and cloud projects hand in hand because just focusing on cloud without innovating at the application layer would leave enterprises with incomplete architected solutions.[4] I also thought that consulting firms would at the very least force the hand of the customers by warning that full cloud migrations could not be materialized with modern (read: mobile) experiences. In other words I imaged that company pain points associated with legacy apps and the highlighting by consulting companies of mobile as a solution would converge and force the rise of unified mobile and cloud initiatives. Companies turned out to be more conservative and usually focused on one then the other, not both at the same time; if they did go mobile first cloud architectures would follow, but not always vice versa.[5]
In 2017 I then shifted gears away from mobile, focusing on the advent of AI, the prodigal child of cloud computing and data, and the business implications of that technology. I had anticipated in 2017 that there would be consolidation in the agency space because of the downwind pressure from customers to work with external partners who were fluent in cloud computing technologies, and soon, AI. Not all would survive as they didn’t have the skill set to do so. My thesis was proven correct. Accenture has made a series of acquisitions (Altima, Octo, Clearhead); Wipro has done the same with DesignIT; SapientNitro consolidated with SapientRazorfish in 2016 and two years later folded into Publicis. WPP has struggled, now bundling VML and Wunderman’s digital fluency with their media consulting offers to bridge the creative and technology consulting gap. In addition, agencies like Huge & RGA have encroached in the space once dominated by the Global System Integrator behemoths, positioning their experience driven approach makes them more nimble than traditional players. In hindsight, I was right for the wrong reasons. Yes, part of the reason agencies consolidated has to do with my thesis that pressure from their customers created a Darwinian struggle where the fittest (those with modern and emerging technologies competencies) would survive.
But, there was an additional force that I failed to uncover, and that was the rise of the CMO/CPO power dynamic within the enterprise. These two poles of power and sources of budget only grew over time and exerted their influence in making purchasing and technology decisions in 2018 and more so will in 2019. For context, previously IT decided what and when software was purchased. The rise of SaaS, the growing need to be nimble, and the business requirement to cater to ever rising and intransigent user expectations (mobile made everyone expect that content should be consumable from anywhere and at anytime with high fidelity) meant that digital marketing and customer acquisitions/retainment strategies became CXO initiatives. While CTO/CIOs pairs working with System Integrators (SIs) carefully planned out the security ramifications of any technology investment, CMOs and CPOs would in contrast call up their digital agency partners to quickly spin up proof of concept use cases for mobile and eventually AI as a way to change business models or conceptualize new ways the enterprise could dialog digitally with users. They were focused on customer experiences while SIs were still focused on the back end story. As a result those agencies started winning more business away from SIs and began to build truer enterprise capabilities, marrying front and back end development capabilities (this calls to mind the tension I evoked earlier in regards to modernizing the back end of applications but not the front end with mobile). This paradigm shift is now best seen perhaps in the way Microsoft is now catering to digital agencies — a group they had ignored previously. These firms are known for their privileged relationships with line of business owners and unit leaders rather than the office of the CTO, and as a consequence, became well versed in the concept of “digital transformation” much sooner than their SI counterparts. Additionally, they tittered between AWS and GCP, but rarely Azure, even though the Azure was more predominantly adopted by the customers they served (think retail). AWS is seen as the IaaS golden standard, so their pole position for infrastructure modernization makes sense, while GCP is viewed as the “data, analytics, and AI” platform of choice. Strategically minded Microsoft has thus pivoted their marketing and positioning to move adjacent to their productivity apps (Office365) to take over mindshare in that last arena subsequently (call it the “PaaS” blue ocean to use MBA speak). They’ve subsequently become more active in modern day agency guilds like the Society of Digital Agencies (SODA) to take their new marketing message to those that have privileged advisory positions to the customer they wish to influence.
In other words, the advent of buzz around AI and data, coupled with a trying need for the enterprise to be more nimble, increased competition in the consulting space; while agencies gained more power, the laggard traditional SIs died off, and the larger firms’ knee jerk reaction was to consolidate or gobble up agencies with consumer DNA to compete and hold off the next generation of consultants.[6] Were it not for the rise of cloud, its convergence with mobile, and their subsequent child AI, these firms would not have claimed more power as they were ahead of the curve in compared to the competition. Like Wayne Gretsky, they had skated to where the puck was going instead of sitting in the goalie box defensively.
Data and AI: Duke & Duchess of 2018, King and Queen of 2019
The other argument that I posited — that data is king — proved to be true as well since I argued that AI would come about from the treasure troves of data being analyzed cheaply thanks to the economies of scale brought about by the cloud. In my 2017 mindset, data would be the greatest asset and differentiator for any organization, if and only if they knew how to use it. They would first have to create data pipelines to both structure data and gleam insights that could be acted upon. The organization that could collect the most data and simultaneously put up moats against their competitors to preclude them from accessing that same gold mine would stand above the rest. In parallel, I implied that they could not do this alone — their internal teams would need be to retrained to become data scientists (“data wranglers”), all while their trusted external partners would have to build their own competencies to guide them through this journey. Indeed,
I came across recently a Boston Consulting Group’s study whose findings lend credibility to this thesis. In a nutshell, they interviewed 1000 execs and managers in automotive, CPG, energy, engineered products, and health industries, with the end goal in mind to get a pulse on their perspective on the current state of AI adoption and the challenges their enterprises face. The resulting study, entitled AI is the Factory of the Future,[7] denoted that 87% of study participants said they plan to implement AI. I had echoed that in the 2017, arguing that firms need to start moving away from asking the questions about their business, “what happened?”, to, “what should we do in the future?” Artificial Intelligence in all its might could drive that type of insight as it enables companies to not be reactive to their business insights but rather proactive.[8] Nothing surprising there — my argument wasn’t that original, nor does it come to a surprise to anyone staying abreast of latest VC investment trends whereby capital is flooding to data & AI first startups.
What I failed to capture however was the immaturity of real applications and derived use cases for the application of AI. This finding is also reflected in the BCG study as well. In fact, while 28% of the correspondents had established a comprehensive roadmap, 72% lacked a detail plain, and only 32% were testing applications of AI. The only unanimously agreed upon use case? Improving operations with AI. In so many words, the study participants concurred that operations as a company prerogative would most likely be affected by AI in contrast to engineering, procurement, supply chain management[9]. There was also substantive more detail around how AI would modernize the concept of factories (see below image), but that is a subject that has been detailed in the media quite in depth for some time now.
The depth of use cases for AI documented have been seemed to be quite shallow, and I would argue that potentially has to do with the fact that organizations don’t have the correct elements in place. I would argue that the key principles missing are based on three pillars: (1) people, (2) processes, and (3) products. Ultimately people drive organizations forward, but to get people together processes need to be established, all while balancing order without putting so much red tape in place that everything gets hamstrung. In a world where there are increasingly more SaaS companies touting their product as groundbreaking, innovative, dare we say “disruptive,” finding the right ones for the right reason to solve a primordial challenge is akin to finding a needle in a haystack. Who do you turn to? And where do you look? And without the right product, an organization can’t leverage its people to put in motion a set of processes to change the status quo and adopt cutting edge technologies to stay competitive.
The BCG study echoes that sentiment, touting that structurally organizations lack four key components in order to successfully adopt AI technologies. The first, as we’ve seen, is roadmap and strategy. This is the “ideation” phase of what applications will leverage or be powered by AI. It’s easy to say “We Want AI” but harder still to figure out what that looks like. Building a business case why doing so matters is even more difficult especially if the bottom line business initiative is in the minds of decision makers at best a foggy concept and at worst a pipe dream. Second, current governance models don’t lend themselves well to implementing AI powered solutions, so those too must evolve for the corporate environment to be friendly to such radical change. For instance, there are (rightly) notions that stipulate AI will cannibalize the jobs of others — if the entire organization top down and bottom up don’t feel invested in the success of the project, then it will fail from the get-go. Intuitive point: if there isn’t alignment on which groups will be responsible for specific initiatives — hence why the roadmap is so important — then the endeavor fails before it starts. This leads us to the third component: companies don’t have the existing skills to fully take advantage of AI as a concept. New programming, data management, and analytics skill sets will need to be developed in house or leveraged externally. So if a certain part of the organization isn’t on board, or rather yet, feels like their jobs are at risk because of adoption of AI powered projects, then these groups need to be heard, empathized with, but ultimately, retrained or taught complementary skill sets. What those skill sets look like ultimately depends on what product gets implemented — which takes us to the fourth missing element: “processes and technologies.” As I had mentioned, finding the right product is difficult at best and near impossible at worst if the business driver is not identified first. Sourcing those external partnerships is akin to sailing a boat without a main sheet if the executive team does not know where the wind is coming from nor what direction they want to navigate to.
People, Processes, Products: Recipe for AI adoption in 2020
Looking forward, I think that by 2020, for AI to be predominantly consumed and readily available, organization leaders need to reframe their mindset by building external partnerships with vendors and consulting firms. First, let’s touch on the mindset change. I first think vendors of AI solutions need to accelerate their investments in product and solution marketing. We need more use cases other than the myriad consumer applications leveraging Face Recognition software. This isn’t to diminish some of the work that has already been done — Google and Microsoft’s approach to promoting iOT solutions as a whole for instance, rather than stitching its constituent workloads, is laudable. More empathetic communication, with less buzzwords and technical jargon, and more vernacular around business benefits has to be taken to market. Selling AI is not the same as selling infrastructure — this is an exercise in ideating and explaining the benefits of new applications. It’s akin to a contractor trying to sell the value of a beautifully interior decorated living room when they’ve only up to that point sold the value of re-doing the plumbing and house walls in granite and not wood — it requires an entirely different expertise! Ultimately those with the most empathy for challenges that can be solved with applications, not infra modernization, can more effectively communicate how to solve them. To complement this exercise in marketing, more technical pre-sales and customer facing engineers — at both software vendors and consulting firms alike — need to be trained to take to market technical concepts to audiences that are keen to learn but may not be familiar with the concepts that will be evoked in their discussion.
Second, the partnership model between customers and their external partners need to evolve. As I had illustrated, the consolidation has resulted in a few organizations with the right skill sets to help with AI technology adoption (since they started making investments there earlier on than those that have died out). Consequently, these few leaders can train their customers’ teams and share processes they’ve developed. Moreover, they can promote the right products that two years ago didn’t exist because they also have developed their own relationships with software vendors (startups and entrenched players alike). We need to develop more knowledge centers internal to these firms to create more competition in the field. More competition in turn will force these firms to differentiate themselves by ideating new offerings that address use cases not previously thought of. If a consulting firm’s métier is to indeed be subject matter experts, they will have to further discipline that in respective verticals (retail vs healthcare vs manufacturing) and use cases (consumer apps, infra “lift and shine” leveraging emerging technologies like AI, devops optimization, digital commerce experiences, etc). That increased rigor will only benefit customers as the use cases will be better flushed out and the associated impact to their customers’ bottom line better anticipated and communicated.
Moreover, firms need to close the feedback look and became interlocked with the startups and larger vendors trying to cater to the enterprise customers they are selling to. Why? Because they’ve established long relationships with those same customers and consequently have a deeper empathy for where they’ve been and perhaps where they should be going. They are usually better trained in diagnosing specific pain points and coming up with a professional services offer bespoke for the customer, unlike software vendors whose sales teams more often than not have been trained to target those same firms with specific scripts in mind. A better partnership needs to evolve between both parties because the former (the software vendor) is a product and not necessarily a process expert. The later (consulting partner) is a process but not necessarily a product expert. Transfer of knowledge needs to happen between both so each can leverage the other’s strengths to deploy a solution that can fit within the confines of an organization’s internal processes and people investments. The consulting company can train their customers, evolve their processes, while the vendor can dive deep and share best practices around how extensible and customizable their product can be. The strategy and roadmap can be built together because two experts are stronger together than separate– this seems trite and corny, but the human partnership will drive the success of a seemingly non-human technology meant to automate and streamline efficiencies.
I believe that in 2018 software started to evolve as a catalyst of new applications enriched by artificial intelligence workloads. We are approaching in 2019 the toddler years of its impact, but for the technology to reach adolescence in 2020, we have to work backwards from the impact of the bottom line it will have on businesses before we start promoting half-baked ideas and concepts. Too often we forget the human element of “digital transformation” and focus too much on the technological. As the complexity of solutions emerge, so will the number of stakeholders solicited for this forward thinking approach to uprooting and evolving their businesses. In of itself that means identifying the use cases for their customers; speaking about AI intelligently and not simply as a concept (which I myself struggle and did so on a few occasions here); and lastly, delivering in time the growth mindset that their customers desperately need in 2019. That is ultimately the last point and the most salient argument that I hope proves to be prescient for 2020: that if we really are to use AI more predominantly, enterprises and their partners need to flex their creative muscle to figure out where and how they’ll implement it. History will tell us in due time if that was the case.
— — -
[1] Source: https://www.bcg.com/publications/2018/artificial-intelligence-factory-future.aspx
[2] Source: http://monsieurpopp.tumblr.com/post/152629160789/mobile-with-a-hint-of-clouds-how-to-pair-a-cloud
[3] Source: http://monsieurpopp.tumblr.com/post/161567081269/where-art-thy-data-brother-the-digital
[4] Ironic that I would say that despite the fact that I had written “the highest growth in cloud computing will come from cloud system infrastructure services, which is projected to grow 38.4%.”
[5] Cloud migration process theory (see New Relic’s set of webinars on their blog for reference) stipulate that organizations should step ladder their plans. First, any net new services should be built with a cloud first infrastructure; then, all non-critical tools should be migrated over as well. This means that the more critical ones and re-writes need to be architected last. For all these use cases mobile is seen as best as an add-on and worst unnecessary. Yes, certain use cases — in banking and retail for instance, where mobile is a do or die situation — will require from the get go that mobile footprint, but it is not mission critical for instance for internal productivity apps. As least that’s the perception. In other words I overstretched the strategic nature of mobile in some use cases — I didn’t appreciate the inherent challenges in infra refactoring.
[6] A similar dynamic is happening within the management consultancy world. McKinsey and BCG are trying to build out their consultancy offer — where they offer “grey matter” and knowledge base strategy advice — with delivery services. I wrote on the matter a while back: https://www.quora.com/What-is-the-future-of-Boston-Consulting-Group-in-the-next-five-years.
[7] Source: ttps://www.bcg.com/publications/2018/artificial-intelligence-factory-future.aspx
[8] “AI, the prodigal child of the marriage between “cloud” and “data”, can help businesses answer that last question… [businesses have] moved from descriptive analytics (“what happened?”), to diagnostic (“why did it happen?”), and now to predictive (“what will happen?”). The next frontier is the promise of AI: the ability to be prescriptive (“How can we make it happen?”)”
[9] Perhaps the data size of the pool of participants is not statistically significant? For the other verticals evoked can certainly be addressed by AI systems, reflecting the participants lack of depth in the subject matter. You can imagine a conversational bot that manages procurement cycles for contracts under a specific threshold, and for more substantial ones automatically punts them to a real human. This consequently reduces costs so only high ticket items get rerouted to a human. But I digress.
