Technology executives in medium and large companies—especially companies who have been around a while and whose core business is not technology itself—often face a daunting landscape. Unless you are one of the few tech execs that have built a large organization from the ground up, or you work for an ultra-modern tech shop, you've probably inherited a “work in progress” or been given a mandate for a significant IT transformation. It may have been the reason you were hired.
Your organization may have been built via a series of mergers and acquisitions, and may contain a mix of bought/built/integrated/on-premises and SaaS products. Some were likely created by vendors and maintained by your staff, others may have been built in house and are now maintained by vendors (often offshore for cost reasons). Your staff may be a mix of contract and full-time in various locations, and often those locations are not due to a cohesive corporate site strategy but are an artifact of acquisitive activity.
The technology landscape may include legacy business systems, including mainframes and 90's-era client server systems; web-based systems from various eras; and probably mobile apps. Perhaps the back-end systems were founded on a Web Services SOA platform from the early 2000s that morphed into a new micro-services platform that was supposed to replace everything else, but was never completed due to budget constraints—and the reality is that most of your customers are still running in the legacy world. And what about cloud migrations? Perhaps you inherited a half-complete lift-and-shift to the cloud, and your CFO now wants to know why your AWS bill is greater than the legacy data center you migrated from.
From a talent and practices perspective, how are your Agile (and other) transformations going? Perhaps you have teams that embraced DevOps and mastered continuous delivery. You would love all your teams to be two-pizza-sized and staffed with T-shaped people. Perhaps you have some of these teams and they are doing great, but you may be struggling to hire more of these types; talent is scarce and the people you want to hire don't want to work for your organization. Perhaps that's because your office location is not in some tech-sexy urban area that attracts millennials, or you are not known as a best place to work, hot startup, or brand name tech company.
Against this corporate backdrop, it should come as no surprise that your CEO and probably your Board have been asking about AI, and perhaps demanding to know your strategy and see your roadmap. Your organization produces and owns a lot of data and you sense that there may be significant opportunities for automation and machine learning in your world, but you've been so busy that you only have a basic understanding of what machine learning is really all about—like Risto Siilasmaa, Chairman of Nokia, last year (Risto Siilasmaa on machine learning). So you prioritize time to familiarize yourself with the space and hire your first data scientist, who costs you a fortune and runs experiments in R, but all you seem to get out of it is a word cloud of the most commonly searched-for terms on your website. Then you read this article on Hackernoon and wonder if you've hired the wrong type of person. Meanwhile, your CEO and Board are looking for answers, and your marketing people have already told the world that you are “powered by AI.”
So you've got to do something credible with AI, and time is running out. Your head is spinning with new tech concepts and acronyms from your recent journey...
Reviewing your options, you come up with three possible ways forward:
- Build your AI capability exclusively in-house. This option is appealing because you know you will need this capability in the long-term, but the problem is time. You had a hard enough time finding a single data scientist; you'll also need data engineers and expertise in AI ops, tools, and infrastructure.
- Acquire an AI startup. This option is tempting and would get you a ready-made skilled AI team, but the problems again are time (lengthy due diligence), geographic location headaches, integration, cultural assimilation, earn outs, etc.—and besides, the market is so hot right now that valuations are sky-high.
- Outsource your AI needs. You do some research and quickly come to the conclusion that some form of outsourced AI engagement will be needed if you want to get ahead and have a credible strategy for your organization. At the same time, you would like to use the opportunity to up-skill your existing team and build your in-house capability for the longer term.
Having done many vendor searches in the past, you have seen a lot of glossy PowerPoint presentations with impressive numbers from large established consulting firms with household names, but often times the results have been disappointing: engagements are time-consuming to set up, and rapid feedback and early results are slow in the making. Quite often, lengthy feasibility studies are recommended as the first deliverables, but this makes you realize that you need a strategy along with an implementation vendor, and some kind of methodology to guide the way. Strategy is definitely something you would prefer not to outsource; after all, who knows your business better than you?
Your experience with Agile and DevOps has taught you the benefits of a highly transparent, automated, disciplined approach to implementation—and also that close collaboration and active engagement are key elements to long-term sustainability after the initial project is completed by your chosen consulting partner. From your research, you understand the potential complexities that machine learning can introduce, and that machine learning is hard, requiring specialist skills and methodologies. Ideally, you would like an approach that captures the best aspects of the Agile, DevOps, and Lean Startup, plus additional best practices from big data initiatives. All this, combined with up-to-date expertise derived from real-world experience with AI projects “in the wild”.
At Manifold, we understand the position you're in, and you're not alone. We developed—and are continuously refining—our Lean AI methodology to solve exactly the kinds of challenges you're facing. The six steps of Lean AI combine aspects of human-centered design with aspects of Lean Startup, Agile Software Development, and CRISP-DM, a methodology pioneered by the data-mining community. Our expert team has successfully used the Lean AI approach across a wide array of industries.
If you're looking to accelerate your data and machine learning initiatives by partnering and up-skilling; if you need a trusted advisor to help you cut through the clutter of available tools and solutions to get to an action plan and begin seeing results; or if you are struggling to lift and shift an aging data architecture in order to take advantage of modern software services and analytics—then we'd love to start a conversation.
When most people think of AI, they think of the areas of technology that are frequently in the news for pushing the envelope, such as autonomous vehicles or robo-traders—but in many cases the greatest business ROI can be found in areas that are less highly visible but often very painful. Many are back-office processes, where much of the hard work involves complex data engineering. At Manifold, we live and breathe applying leading edge AI in the business world, rather than experiments in the ivory tower. We focus on cutting-edge techniques that garner tangible value, rather than hype and headlines. And we're here to help you accelerate your business.