Artificial intelligence (AI) hasn’t left the news headlines since ChatGPT exploded into mainstream use. Since November 2022, when the game-changing chatbot was launched, AI has moved from the image of stealing jobs to being a productivity enabler for all workers. In our own business at Altron, AI has been able to identify non-functional cameras in our Netstar logistics offerings, ensure spare parts are available for our retail POS clients and process semi-structured documents such as bank statements into our transactional systems.
However, launching an AI program isn’t going to deliver guaranteed results without serious planning. As with any technology, the delivery of AI solutions needs to be built within the structure of a strategy that provides guidelines for the organisation. To this end, Altron has developed five crucial steps (with the acronym of RAIdED) to help you initiate a successful and growth-focused AI initiative.
Step 1 – Research. If you want to launch an AI strategy within your organisation, first understand what is possible with AI, especially within your industry. There are literally hundreds of AI websites, podcasts, books and articles you can learn from, with a primary goal of understanding what is possible in AI, and also to catalogue use cases relevant to your business. I can recommend medium.com, the AI Podcast and dailyzaps.com, but there are many more.
You’ll also want to look at which technologies fit your needs in this step – Microsoft, Amazon and Google all have industry-leading AI services available in their cloud solutions, and there are specialist players too, such as Datarobot and IBM’s Watson.
Step 2: Aspiration. What is the short, medium and long-term AI vision for your organisation? Gartner talks about defending your position (looking for quick productivity wins, such as CoPilots), extending your position (building custom solutions that create differentiation from your competition), and upending your position (focusing on a big disruption play). Their position is that most organisations would be looking at the middle option – extend – as their long-term play.
Altron’s experience is that most companies will want to take a hybrid approach with regard to their aspirations, and we represent this with the below matrix.
AI impact matrix with examples, Paul Morgan, Altron.
The matrix indicates that although employee productivity solutions, such as deploying copilots, are often seen as a relatively easy place to start, it is often difficult to measure the real return on investment.
Step 3: Ideation. Brainstorming with your team and other areas of the business, or activities like hackathons, are good places to start. Collaboratively identifying business problems will translate into a pipeline of high impact and feasable projects. Focusing on the impact of potential projects will ensure that the organisation prioritises its AI investments and direction. For serious buy-in, conversations with leadership will be more fruitful if some projects with hard-impact measures, such as cost reduction and revenue acceleration, are selected.
Step 4: Ethics. Before actually starting any project, or even a pilot, the question needs to be asked – should the business be building this? Tackle your AI with a mini-governance mindset to unpack potential problems before they become actual problems. This will minimise risk while improving long-term outcomes and sustainability. KPMG has a useful mini-governance layer that can be expanded in your AI strategy, and also run as an ethics gate before starting any development. According to KPMG, the questions that should be asked are: Is the solution fair? Does it have resilience and integrity? And is it explainable? If these can be answered positively and truthfully, and the solution has an identifiable business value, then proceed to the next step.
Step 5: Development. To ensure long-term success, organisations need to invest in training and upskilling internal staff and potentially partnering with trusted service providers. For each project, we suggest an initial sprint to identify if that solution really will be feasible and impactful. If expectations don’t materialise, then don’t be afraid to drop that idea and move onto another one. But first make sure you have clearly identified, and documented, where failure occurred. Continual improvement is critical.
AI applications need significant effort to ensure they continue to deliver at scale. Looking at MLOps to retrain models and ensuring they continue to provide the business value expected is essential.
This approach may seem daunting, but Altron has a trusted multi-technology AI practice and can assist at each stage. Talk to us about how AI can transform the future of your business today.
Share