The struggles and potential of AI integration in modern businesses today

It may not be a surprise due to the nature of how rapidly generative AI (Gen AI) evolves, and there are obvious challenges for traditional business to transform with unknowns.

With the magnitude of potential impact, it is though still worrying to notice that after insight into many businesses that many struggle to progress beyond the minimum viable product or proof of concept stage in their AI initiatives. These efforts often stall due to being siloed, not part of core business, and prompts without high quality.

It is dangerous, as AI will have profound transformational impact. It is not a hype. The deception may be related to the the misconception that the technology is not mature, but it is.

Meanwhile, the real issue was that failure in vision, planning and execution on how to engage the organisation, and how to incorporate feedback loops for fine-tuning, reinforcing learnings, and flexibility to adjust in the fast-evolving landscape of tools.

Generative AI should be infused into core business processes rather than treated as an add-on. There is a choice and the consequence of business outcomes will be very different.

How to infuse AI into your business

To infuse AI into the business requires the traditional buy or build choice. Any organisation needs to reason for about where to build and where to buy solutions where AI is implemented when implementing AI in collaboration with a partner. It's crucial to select the right partner for AI infusion in specific areas, such as GitHub Co-pilot, Salesforce CRM, or Power-BI Co-pilot—areas where vendors have a strong capability and even stronger roadmap. However, even with the right partners, cultural adoption and challenges remain, including fostering curiosity, encouraging the change, patience and playful exploration of new ways of working.

For me personally, GitHub Co-pilot has been a great experience from day one. However, the Office 365 Co-pilot experience wasn’t that impressive to start with but just 6 six months later it shows another side, and I can clearly see how profound it evolves as a continuous improvement journey.

My main reason for this short blog post though is to focus on observations how many business companies struggle into to evolving their core business with help of AI. This involves not just adopting Generative AI technologies but embedding them deeply into their core operations to drive meaningful transformation. Many organizations focus more on acquiring tools and technologies rather than fostering a culture of AI adoption.

To reinvent the customer engagement, an effective AI infusion requires a shift in the mindset, processes, and to be bold.

A positive observation though is that there is no lack of ideas. Some businesses invite consultants, or run internal workshops and, after a few weeks or months, end up with a list of 40 key ideas. This creates a sense of accomplishment. However, some of these initiatives starts but fail due to underestimating the complexity involved in adopting AI.

Additionally, these efforts are often peripheral, not integrated into the core business with the core people. This approach can divide scared resources, polarizes organizations, be a bottleneck for AI adoption as impressions get negative and dilutes focus.

Why is it important to infuse AI into your core business?

Customer experience is likely to change rapidly in many industries, with co-pilot drive driving command-line interfaces and voice interaction reinvent the ways of workingthe ways of working will be reinvented. There will be a shift in customer experience and engagement. The effect in the core business areas will be the largest.

In a fast evolving landscape, what should be our benchmark or inspiration? It is not easy to keep up. Change is faster than ever, but will never be this slow again and AI accelerates this even further. One example that we can gain public inspiration from, is how Open AI evolved so greatly in AI driven Phyton coding through Open AI Codex program – which is today the corner stone of GitHub Go-Pilot.

Python developers at OpenAI play a crucial role in various aspects of the AI development, and they were both hired and consulted. Still human in the loop.,

Meanwhile, Gen AI infused a generational shift in how our companies can compete, by increasinge productivity and way of working. Today, it is highly automated to write Phyton code, and in the mid term the role of a developer will change as you can generate application by being good to know what you want through prompting.

In all large companies, a lot of knowledge exists and partly digitalised but mainly in the head of the people within the organisation. In traditional business, a vast amount of human competence will be hard to replace when employees retire. To transform within the core business, these people will be a core asset for modelling, quality assurance, but also as well as fine-tuning and reinforcing learnings. AI infused services built from core business knowledge has strong potential to unleash a lot of time for human creativity and business growth.

Learning from successful examples should be the guide. There is a lot to learn here, and it is evolving fast. In addition, what it takes in focus and resources tosuccess requires focus and resources, succeed and it means stopping the practice of doing many things at the same time.

Finally, there are no short-cuts. You can’t train for a marathon by watching TV.

Companies excelling at AI infusion

A selection of organizations are excelling in AI infusion, and the results are impressive. With strategic management and motivated teams, these companies achieve rapid advancements through modern AI infusions with unknowns. Here’s what they do differently:

• Strategic integration: They identify key areas where AI can transform core business processes and embed AI deeply into these functions. This leads to significant operational improvements and efficiencies. They are not afraid to change process, and work with change management. Perfect use cases are customer assistance to suggest, recommend and verify, co-pilot to add value to product or service, or where unstructured data knowledge can deliver business value through interaction via natural language prompts and follow-ups.
• Continuous improvement: Strong support and feedback loops within the organization and with customers allow for constant fine-tuning and learning. This ensures AI solutions remain effective and relevant.
• Leadership support: Active management involvement fosters a culture of innovation and dedication. Leadership motivates teams and aligns AI initiatives with broader business goals. It helps to understand how to either increase growth or be more efficient.
• Robust learning and adaptation: These organizations generate insights that create robustness and competitive advantages in the fast-evolving AI landscape. They adapt quickly to new developments, continuously enhancing their AI capabilities.

Key elements of company Gen AI success

Key elements of their success include starting small with focused initiatives, using robust AI platforms like OpenAI and Microsoft AI Services, maintaining a flexible architecture, and implementing strong privacy and security measures. These practices not only drive fast, impactful results but, also build a sustainable foundation for ongoing AI-driven innovation.

Even for leading companies advanced in developing digital twins of their core business in traditional industry, these are often separate from Gen AI initiatives. Gen AI has the opportunity to streamline digital-twin deployment, while digital twins could refine and validate Gengenerative AI output.

The transformation journey from cloud into digital to AI is accelerating and it will just go faster.

However, I am concerned that many companies do not see AI as a strategic asset to be infused into their core operations and decision-making processes. Without this perspective, they are merely testing the waters rather than fully diving in. Gen AI will allow a great opportunity to unleash our internal and customer creativity. Let’s jump into the water and spend less time on peripheral and visionary abstract creation. We have the tools at hand to bring intelligence to our core business.

What you don’t need

• Your own AI model and system
• Data scientists as isolated team
• No interest from top management, and board members
• Visionary 40 future use cases with naive ability to focus and with low change of business outcome
• Be far away from user impact and real time learnings
• A lot of dependencies outside of team control
• Wait for the perfect level of quality data

What you need (in a larger org there is no limited to simplify)

• A key space to operate within and that can transform core business
• Cross functional software team that harness human and digital knowledge, and with ability to fine tune, reinforce learning and assure output quality (your business context prompt engineer, data quality engineer, domain engineer, curious modern tech software engineers)
• Established support and feedback loop with larger organisation
• Start small and stepwise expand (true value will come over time)
• Select a Gen AI platform: Open AI, Microsoft AI Services, AWS Bedrock, Google Gemini (sorry, there are no EU based company in this space that can compare)
• Time to play and adopt flexible architecture (quickly evolving landscape)
• Management attention
• Independently, if it is an internal or external customer: be close and adopt by learning

What you need to incorporate (see European Artificial Intelligence Act)

• Privacy, terms and cybersecurity (compliance, IAM, encryption etc)
• A MVP process, which most likely should start with the most advanced model available, and then fine tune and optimize based on 1000s of models available
• Mix Large Language Model with Small Language Model (best fit for optimal use, and cost)
• Mix with Semantic Kernel or LangChain to interface products and core systems
• A mindset that you may rethink your prompt capability in the next 6 months

Related items you may like:

Generative AI for shop floor transformation

AI for manufacturing