{"id":17844,"date":"2019-11-11T09:00:59","date_gmt":"2019-11-11T08:00:59","guid":{"rendered":"https:\/\/www.microsoft.com\/en-gb\/industry\/blog\/?p=17844"},"modified":"2020-07-31T14:56:44","modified_gmt":"2020-07-31T13:56:44","slug":"implement-ai-mid-sized-company","status":"publish","type":"post","link":"https:\/\/www.microsoft.com\/en-gb\/industry\/blog\/cross-industry\/2019\/11\/11\/implement-ai-mid-sized-company\/","title":{"rendered":"How to implement AI as a mid-sized company: 5 practical steps to success"},"content":{"rendered":"
48% of businesses are currently experimenting with AI technologies \u2013 as revealed in Microsoft\u2019s recent report, \u2018Accelerating competitive advantage with AI<\/a>\u2019. That means, for the majority of organisations, particularly mid-sized businesses, the need to implement AI simply isn\u2019t on their radar.<\/p>\n In truth, it\u2019s likely that companies like this didn\u2019t realise that the internet was a \u2018must do\u2019 at the turn of the millennium either. That\u2019s why it\u2019s essential to help these sorts of businesses understand not just the importance of technologies like this, but how they can integrate them across the organisation.<\/p>\n <\/p>\n In my experience, there are three main reasons why AI opportunities can be difficult to see.<\/p>\n The trick to successful implementation of AI is to refuse to over-commit at the early stages \u2013 this, instead, should be the exploratory stage where a business discovers AI\u2019s impact, and how it should be deployed.<\/p>\n Once the impacts are fully understood it’s time to develop a roadmap, incorporating each part of the business that will use or be affected by the technology. This leaves room for smaller initiatives to be drawn up and integrated into the \u2018big picture\u2019 programme.<\/p>\n Oversight is also critical to success. The team in charge of executing the AI master plan needs to engage with the rest of the business. That way, they can evaluate success (or potential failure), serve additional knowledge, and give the leadership team necessary feedback.<\/p>\n With this in mind, I believe there are five core steps to making AI integration a success.<\/p>\n <\/p>\n Gaining internal buy-in at the earliest possible stage is vital \u2013 this starts with the leadership team. AI is not just another IT project, after all. You should then look at which areas of the business will also be benefiting from the technology and get them involved in the process. It won\u2019t be long until AI is essential for many operations across a business. Don\u2019t get left behind.<\/p>\n Appoint someone to own your AI project. This should, ideally, be someone with strong leadership skills, as they\u2019ll be heading up a cross-functional team and informing company leadership of its progress. Your \u2018project owner\u2019 should also be considering the skillsets and expertise needed to bring the project to completion, whether internally or with your partners.<\/p>\n It\u2019s their job to drive through success. Everything must be measured and kept in control. It\u2019s the only way to dodge those high failure rates.<\/p>\n <\/p>\n Depending on the size of your company, this is the ideal moment to construct a multi-disciplinary team. Keep this to around three people. It\u2019s likely you\u2019ll need to gain a fundamental understanding of AI, the opportunities, and the challenges. This overview helps guide you as you build out the required capabilities of AI and your proposed solutions.<\/p>\n Seek out partners with experience of implementing data and AI projects for similarly sized businesses. These partners will have a good understanding on the challenges you may face, and how to circumvent them. As with any IT project, but especially AI implementation, the sooner you get them on board, the easier it will be.<\/p>\n <\/p>\n Why do you want to implement AI and what do you plan to do with it? Begin by building a solid business case, focusing on key challenges, and how AI can help overcome them.<\/p>\n If AI can\u2019t solve these problems, look at where your organisation could become more efficient through automation, instead. For many smaller companies, this will prove an excellent stepping stone to future AI adoption.<\/p>\n Indeed, given failure is linked to a project\u2019s overcomplexity, it\u2019s worth considering whether your first AI project could be used to automate a simple process. You don\u2019t have to re-build your business from the ground up with an AI foundation. It\u2019s all about giving you time back to focus on more critical tasks.<\/p>\n I recently worked alongside suit e-tailer The Drop<\/a>. During that time, we identified three areas where AI could benefit business process, while improving the experience for customers.<\/p>\n For The Drop, AI helped them streamline the supply chain. This resulted in faster deliveries and fewer returns, since measurements were accurate and mistakes were found before production.<\/p>\n <\/p>\n AI is only ever as good as the data it has. So, once you\u2019ve identified how AI will assist you, you must then look at what data you have or need to ensure it works properly. Your data should be clean, organised, and easily accessible for the technology.<\/p>\n A word of warning, though. It\u2019s possible your IT team may push back on this request \u2013 but the data is owned by the entire business, not a select department. As long as that data is secure, flexible access is a must to capitalise on the value it holds.<\/p>\n The Wild Me project<\/a> is a good example of using data \u2013 specifically, images \u2013 for the greater good. Anyone is free to upload animal images to the Wildbook Cloud, where they can be catalogued and tracked. That crowd-sourced data then helps scientists make informed decisions over conservation efforts.<\/p>\n <\/p>\n With the other four steps complete, the final step is simple. With your business identifying a business challenge, how AI can solve it, and the data it needs to do so, you can start rolling out your AI initiative.<\/p>\n Start with a proof-of-concept. It helps you achieve your stated scope and scale of the project. Next, create a model; build and implement AI. Test it. Ensure it\u2019s delivering what you need. You\u2019ll be obtaining results and value in no time \u2013 whether that\u2019s increasing process efficiency, analysing data, or making customer experiences truly personal.<\/p>\n <\/p>\n And I say all this as someone who has been on the journey you\u2019re preparing for \u2013 at the start, we used Microsoft\u2019s tools and products, like the pre-built Cognitive Services<\/a>, and the cloud capabilities of Wirehive<\/a> to build a chatbot. With the initial groundwork done, and the experience gained, we\u2019ve been able to work on even more complex projects, fuelled by increasing amounts of data and machine learning models. If we can do it, so can you. Good luck on your journey.<\/p>\n [msce_cta layout=”image_center” align=”center” linktype=”blue” imageurl=”https:\/\/www.microsoft.com\/en-gb\/industry\/blog\/wp-content\/uploads\/2019\/10\/CTA-image.png” linkurl=”http:\/\/aka.ms\/AcceleratingAI” linkscreenreadertext=”Link to download Microsoft’s AI report” linktext=”Download the full AI report” imageid=”17871″ ][\/msce_cta]<\/p>\nBeing held back<\/h2>\n
\n
Step 1 \u2013 Teach the benefits of AI<\/h2>\n
Step 2 \u2013 Build a team or partner up<\/h2>\n
Step 3 \u2013 Identify the right problems<\/h2>\n
\n
Step 4 \u2013 Get your data ready in advance<\/h2>\n
Step 5 \u2013 Activate with AI<\/h2>\n
Find out more<\/h2>\n