AI

How AI as a Service (AIaaS) helps companies with one of the great innovation headaches

Share this post:

For the best part of a decade now, businesses have dreamt about achieving a radical competitive edge through use of Big Data. In the past years, the field of big data, analytics and AI has been in rapid development. In the field of technology supporting AI and Data Science, the pace has indeed been exponential. Therefore, it is surprising why few businesses are working aggressively with this game-changing opportunity.

In IBM Denmark, we have built an approach to deliver AI value to clients in weeks, based on open source and standards, trust in AI and with a repeatable framework that can accelerate clients’ internal capabilities as well. We call this AI as a Service (AIaaS).

Challenged with death by a 1000 Proof of Concepts

We have developed the AI as a Service framework, as we recognize the need for expertise to realize and prove the value of AI in practice.

Ambitions of competitive edge through AI is often the imperative for businesses in a digital age, where a lot of time and energy have been spent on crafting bold strategies, creating portfolios of ideas and AI concepts in Power Points, while external innovation units or AI units, have been setup to tackle this urgent, but complex task.

Mostly we see that the bold initiatives loose momentum and get overwhelmed by the many unknown challenges when trying to get AI from ideas to practical, differentiating experiences for customers or employees.

The pitfalls are many, often spearheaded by the lack of understanding between those who build AI, and those who are hoping to drive monetary value out of it, i.e. the data scientists and the leaders.

In addition, AI initiatives suffer the death of a 1000 Proof of Concepts without tangible outcomes at the end, and actual success stories to build on in the organization. The gap between strategic AI ambition and actual AI application is still one of the larger ones in today’s digital transformation imperatives.

So how can we avoid AI to be a victim of technology fascination, and instead make it work for the benefit of people and business?

AIaaS – a lean innovation 4 Step Process to AI across your business

 AI as a Service (AIaaS) is IBM’s proven approach to doing that. It is a process that facilitates all aspects of AI innovation. From idea-ting on how AI can provide value to actual, with a scaled implementation across a business as target – with tangible outcomes in weeks, not months.

Our AIaaS process consists of 4 steps:

  1. Mobilize. First step is about mobilizing the right stakeholders and users to decide on where to start with AI in the business. Most organizations spend their time in this first, initial step. In our process, mobilization is a crucial part of aligning expectations and focus that ensures agreement on the right problem statement, as well as questions on availability and quality of data.
  2. Think. In the think phase, our “hands get dirty” with data exploration and modeling. Our data scientists work in sprints to build the first AI model that will create the foundation for an AI Meanwhile, we bring in complementary skills from management consultant, UX Design, and IT architecture, to ensure the AI product is valuable for both the business, IT and end-users. This process provides tangible value with an AI model, business case, and Product Vision & High-Fidelity Prototype in only 4 weeks.
  3. Transform. We transform the AI model to a Minimum Viable Product (MVP). An MVP is the first digital product, good enough to release to selected users for real-world testing. This is delivered to selected users for testing and validation within 8 weeks or less.
  4. Thrive. When AI is proven successful, we make it work across the organization. Thrive is the process of institutionalizing the AI product and its processes for everyone. Sometimes, this it is just a matter of scaling up bandwidth and storage, sometimes AI products are improved, processes adjusted, or a product and roll-out roadmap is built and executed on.

We recognize that AI innovation is a challenging, necessary and exploratory process for our clients, and a function that often is wished to be kept inhouse. However, as a multifaceted, multidisciplinary process, which requires rigid work with several different methods to build a differentiating product, we see that a lot of businesses greatly benefit from experience of experts to start off their own transformation internally.

Initially great creativity is needed, when AI begins with idea-ting concepts through IBM Design Thinking to ensure value for the intended user. Then it is a mathematically complex process when creating the AI itself, either through AI services, machine learning or deep learning. Lastly, it is about orchestrating an agile team can deliver the technology that should deliver delightful experiences to users.

Our track record shows, that the process of AIaaS enables a structured, beneficial way of balancing both competencies of data science, IT and business consulting, as well as balancing the technical delivery with the role of ongoing change management that comes with AI.
Ultimately, working with AI is about shifting your organization to better processes, through use of new technologies, driven by data, mediated by AI models and enabled by compelling user experiences that satisfies needs and wants of employees or customers. And the challenge is, that great AI user experiences requires you to do everything great, all at once.

AIaaS is more than a technical delivery; it is IBM’s approach to help our clients do UX, AI, Data, new technology and change management, all at once.

 Our experience shows that the holistic approach of AI as a Service, decreases risk of AI innovation, while improving time-to-market, product outcomes and value for the business. At the same time, it also provides our clients with a blueprint for AI going forward, thereby accelerating internal know-how and ability to execute.

It also minimizes the commercial barriers of working with AI for our clients, and the agile delivery framework ensures alignment, transparency in creating the AI, while increasing momentum for the AI.

 AIaaS delivers trusted AI for an open world

One element of AI delivery is the process, another is the foundation on which it rests, i.e. the technology. In IBM, we have a differentiating focus on creating AI that is platform agnostic, and explainable and ethically designed.

 As we see a world increasingly abandoning proprietary software licenses, monolithic applications with vendor lock-in, and moving towards an open source and micro service based approach to delivering new digital services with speed, scale and flexibility, our approach to AI is complementing this move in several ways.

  • We are making our Watson AI services available across cloud environments and providers. Regardless of whether you have your use cases or programs running or planned, on a non-IBM cloud, you can still enjoy the leading platform in data & AI that orchestrates your open source tools.  Our view on multi-cloud and a more agnostic IT environment is also supported by IBM’s recent venture with RedHat
  • We also recognize that more mature businesses get operating AI in production; the need for versioning of AI models, continuous improvement of models, consolidated governance on data and models and explanation of AI, is needed. With Watson OpenScale, we are able to do exactly all of that.

 In IBM, we believe that AI should be designed by determined ethical guidelines, based on open standards and formed and enabled by the very people that AI is intended to make a difference for (does your company have an opinion on what ethical AI consists of?).  With AI as a Service, we build on these principles for AI delivery, while accelerating clients’ own capabilities to become a Cognitive Enterprise

Wrapping up, we see our AIaaS model working with practical application of AI because it is about:

  • Based on lean innovation. By now, it should not come as a surprise that methods like Design Thinking and agile provides an edge in decreasing innovation risk, while increasing outcome
  • Demystifying AI by doing it in practice. Too much AI is slide-ware-driven. You can only learn so much about AI, and make decisions about AI, when it is driven by theory. By using lean innovation methods, you are able to demystify AI, learn about AI in a use case context and make better decision about what to do with AI in just weeks. This also brings me to my next point:
  • Being pragmatic about data. Organizations’ messy data or a data transformation programs can get in the way of focusing on extracting value out of it. Do not let it. There is tremendous value to be capture, just working with the data you have at hand, great talent, and new methods.
  • Making AI real in the organization. What is the value of the world’s best prediction model that is not deployed in your organization? Nothing. The barrier of getting the first AI project deployed in your business, is definitely real. AIaaS has a proven method for actually getting that first, crucial model deployed.
  • Driving AI through user experience. What is the value of the world’s best prediction model that is not being used by people? Nothing. If you are not able to tie together your AI efforts with good user experiences solving real people problems, it is not going to make a difference for your intended users.

These five enablers have proven successful to launch AI, and doing so in a way that supports the paradigm of think big, start small, scale fast, which enables your business to prove value by minimum investment, and scale as you gain confidence in your AI project.

This is exactly what we have seen over the last few years in Denmark. There is an increase in businesses going from ideas about AI to tangible AI-driven products and services.  However, the window of opportunity to get a head start and competitive edge with AI has not passed, but the next year will be crucial for businesses to go from ideas of AI to value from AI.

Senior Strategist

More AI stories

Nordic Organizations Now Need More than 11 Times Longer to Close Today’s Skills Gap

Over 120 million workers in the world’s largest 12 economies may need to be retrained or reskilled as a result of AI and intelligent automation in the next three years. Our skills gap – also in the Nordics – is real and it is now taking 12 times longer than four years ago for organizations […]

Continue reading

Self-Learning Data Catalogs – Data Catalogs’ Sequel

Self-Learning Data Catalogs – Data Catalogs’ Sequel This is a sequel to the future progression of data catalogs. How to enhance and automate data management and analyzing processes with Data Catalogs by supporting the development of Artificial Intelligence (AI)? In these development processes, what are the main issues? Introduction I ended my first blog post […]

Continue reading

Journey to AI – An IBM Garage Event

You are invited to our Journey to AI event. “Enterprise AI” is about solving sophisticated business problems in highly dynamic environments. This requires an understanding of well-defined use cases and starting points, and an acknowledgment that the bottleneck now is in management, implementation, and business imagination more than the lack of technology AI is transforming […]

Continue reading