Five Hacks for Using AI in Digital Transformation

Back when I was starting to pen down some ideas for my book, Accelerated Digital Transformation, a voice inside me kept repeating ‘if you are writing a book on transformation, then you should transform the writing of the book’.

One fine day, I got one of the transformative ideas for the writing of the book, to co-author the last chapter of the book with an AI bot. This was well before ChatGPT! It was fascinating to see an AI write some awesome hacks summarizing the essence of the book, in a few paragraphs. I got goose pimples at the wisdom of the bot to have understood my experience through my book and translated it into these insightful hacks.

In the last few months, after over 100 million humans interacted with ChatGPT in less than two months and every interactive dialogue with the AI left them both dumbstruck and inspired, this has become a norm. Humanity just expects AI to respond intelligently to standard queries and interactions. This is the exponential pace at which this technology is taking over our world. However, the enterprise world, as usual, remains sceptical.

The top executives are hoping the AI wave is just another fad and if they can just bury their heads in the sands for long enough, the storm will pass and their enterprise life will bounce back to normal. Well, in my humble opinion, this storm isn’t fading away anytime soon and, to the contrary, AI is the new norm.

For this reason, here are my  Honeycomb Hacks (HH) for AI adoption within the digital transformation journeys of established enterprises:

1. Seek out the passionate

Hidden within every enterprise there are gems who are passionate about new technologies, who are curious and who have a continuous learning mindset. Seek those with a passion for AI and assemble them into a community of passion. Appoint a leader and give them access to the top leadership and attention. There is so much inertia and scepticism within established enterprises, that it takes true passion to break through those walls of resistance.

Please note, I stay away from using the standard term ‘communities of practice’, but instead specifically seek out people with a passion for AI. This self-motivated community will automatically start action around AI, including spreading awareness of AI and its potential within the larger enterprise stakeholder groups. Let the community organically evolve.

2. Wins go viral

Shape one or two initial use cases within the enterprise which are both fit for AI-based transformation, as well as visible in their outcomes. For example, one use case could be around forecasting demand, or predicting pricing in real-time, both of which have a direct revenue impact and are transformable using AI-based deep neural networks.

After identifying the use case and engaging with the relevant executives, ensure that one use case is totally executed. See the use case through, from soup to nuts; from the development of the AI model to its implementation in the live environment, through to clear visibility of business outcomes and then a larger enterprise-wide celebration of the success. Wins have an inherent tendency to go viral. Once one use-case is taken through the enterprise crevices, to execution and outcome, a thousand new flowers of AI will bloom automatically, breaking organizational inertia and creating momentum.

3. AI factory

In parallel to the use case going live and right after its success, the business case for a centralized AI factory needs to be developed and approvals sought for the resources and costs. I am a strong advocate of federating capability within a larger enterprise, as opposed to creating monolithic, central structures.

However, for AI, given the extreme specialization of this field, initially for the first phase of three-plus years, we need to have a strong centralized, factory model for AI. I am telling you, setting up this AI factory is easier said than done. AI capabilities are most sought after globally, perhaps second only to cyber security digital capabilities. The enterprise will need to ensure they have an authentic, transformation-aligned proposition to be able to attract the right AI talent.

The AI factory once formed, needs to develop a library of AI use cases across the enterprise. This portfolio of use cases then needs to be portfolio managed to execution, depending upon the availability of data for the use case, resources available and business impact.

4. Data janitor

You will have come across this cliché many a time: ‘garbage in, garbage out’. In my experience, this cliché is most relevant to delivering the AI promise within established enterprises.

I would conjecture that AI success is 95% contingent upon data. And typically, within established enterprises, the state of data is in a mess. There are multiple silos of the same data, hidden in departmental stores, and not available to anyone else to use. This needs to be fixed before the AI factory can start delivering AI use cases at scale.

You can hack your way for the first use case, to create that right win, however after, the janitorial work over enterprise data has to commence. In my experience, the technology patterns for delivering clean enterprise data are now sorted. You can choose between data lake and data mesh architectures and there are enough cloud-based technologies to help you get the data sorted and make it available for AI algorithms.

The key, however, is to institutionalize as a habit, disciplines relating to data governance, data lineage, metadata management et al. These disciplines ensure that there is no garbage going into your data lake, which will then feed the AI algorithms, which will create magical outcomes for your business.

5. Eyes on the horizon

The last hack I will leave you with is keeping your eyes on the horizon as you develop and hone AI capability within your enterprises. AI is one of the fastest-evolving technology fields today. It is well positioned to become one of the few ‘general purpose technologies’, which will have a profound impact on humanity and our future.

However, given the self-learning nature of the technology, there are significant concerns around the risks it poses, ethical and bias concerns and an entire body of research being conducted on how to regulate and govern AI.

Enterprise leaders need to keep a keen eye on this evolving area of AI governance and ensure adequate controls are proactively developed, so that AI use cases can be widely scaled across the enterprise.

Digital transformation is a team sport and it is fascinating that in the future we will have to team up with both our human and AI colleagues to deliver the promise of digital transformation in an established enterprise. There is, however, a long way to go before enterprises become open enough to allow AI algorithms to join us and contribute to this transformation journey.

I believe the above Honeycomb Hacks can help enterprises lay the manure, where passion for AI and AI capabilities can grow and nurture.