How Can SME CEOs Adopt AI To Create Value?

SME CEOs can create value by adopting AI with this 5 As framework: Agree; Analyse; Assess; Apply; Adopt. Read on to learn how to apply this framework at your business.

DO NOT talk about AI adoption!

The first rule of AI adoption club is: you DO NOT talk about AI adoption!

Ok, so the Fight Club movie reference is pretty lame, but it serves a purpose.

If you are coming at this trying to shoehorn AI into your business because you heard that AI is a total game changer and your FOMO is kicking in, then you're likely to spend a lot of cash achieving not very much beyond adding complexity to your digital landscape.

That's because while AI absolutely can be a game changer, it is rarely a good idea to decide on the solution before first thinking about the problem. This concept is just as true for AI as it was for blockchain, the cloud, big data etc.

For example, using generative AI to automate your customer service function just because ChatGPT is rocketing up the Hype Cycle might not delight your customers in the way you expect. Rather than giving customers faster, always available and accurate support, perhaps you watch your customer satisfaction score plummet because your "intelligent" agent is serving up fast but inaccurate or irrelevant information in an off-brand tone that actually results in more calls to customer service agents than before. Not only are customers unhappy but your support team is busier than ever before. This doesn't mean the tech was bad, more likely you just weren't ready to adopt it, you didn't need it, or you didn't spend long enough thinking about the problem to realise that RAG (retrieval augmented generation) would have been a far better way to run an AI customer service agent if you want to serve up accurate responses.

Perhaps you run a manufacturing company and you are looking to boost efficiency in your warehouses. If you produce goods to a particular standard, following a certified process you probably don't want to employ machine learning to constantly tweak how your inputs are used to make your outputs. Creating a precise, predictable and standards compliant output using simpler, cheaper and more predictable forms of automation is almost certainly a better option. On the other hand, using machine learning to identify defects, or for predictive maintenance, or for identifying process bottlenecks could be highly valuable.

In more general terms, using AI to automate a bad process is likely to create even more noise for your teams to straighten out.

Even if you have a sound problem to solve using AI, is your company ready to adopt AI?

Using AI when your data is low quality can produce inaccurate results.

If your people are not enabled to adopt new AI tools and processes, they won't.

If you get your first AI enablement project wrong, don't be surprised if you are put off using AI while your competitors embrace it.

So How Should SMEs Adopt AI?

When you start thinking about AI adoption for your business, it is easy to jump straight in with questions like:

Is an AI system appropriate for my business?

Is my business ready to adopt an AI system?

What are the risks of adopting AI?

These are all valid questions, but the first questions you should be asking have nothing to do with AI, such as:

How can I deliver the best possible customer experience?

What frustrations are causing high churn in the team?

What bottlenecks do I have in my value streams?

How can I increase the lifetime value of my customers?

Your first questions should be about people. Start with customers then follow closely with employees.

Your next questions should be about your business processes those people use.

Finally, pause to consider the data you acquire going about your business as usual activities. Often data is just a by product of delivering an essential service, but what if you could transform it from a historic record of fact into a predictive powerhouse? Might that have potential to uncover a brand new revenue stream?

This should give you a pretty good idea of the challenges and opportunities facing you, along with the potential value you can create from addressing them.

Now we can start to consider whether AI is appropriate for any of those use cases.

Types of AI

AI systems have become increasingly prevalent in recent years, with proven capabilities across many domains. To help you navigate through the potential options it can be helpful to zoom out and consider some broad categories of AI system:

General classification AI systems
Natural language processing/understanding
Generative AI
Biometrics
Anomaly detection

General Classification AI Systems

If your business has large amounts of input data, such as application forms, fault reports, expense claims etc, a classification AI system could potentially add speed and accuracy to your operations. Classification AI Systems learn to identify patterns and relationships in data and can associate those patterns with automated actions.

Personalised Content / Personalised Communications

A type of classification system that attempts to understand not only who your customers are, but what information they need from you and when they need it in order to take the next action you would like them to take, such as transitioning from researching your product to purchasing it, from using your product superficially to fully engaging with it, from being a happy user to being a brand champion.

This is done by examining patterns in data to classify users into groups and using subsequent data analysis of past customer interactions to determine what is the best content to serve or communication to send to have the most positive impact on the customer's experience in order to move them closer to your next desired action.

Natural Language Processing/Understanding

If your business has a lot of customer touch points creating a high volume of communications, whether that is pre-sale or post-sale, there could be an opportunity to improve customer experience and improve operational efficiency by automating some of those communications. But be cautious, replacing a human touch with a robotic feel that is ineffective can be counter productive. Also consider NLP/NLU if you have to perform processing tasks on large amounts of written or audio data.

Generative AI

If your business produces a lot of text, image and even video output, generative AI could give your teams super powers. GenAI agents can work as a co-pilot to provide inspiration and take the pain out of repetitive tasks such as writing boilerplate code in software applications, taking notes in critical meetings and generating images to bring marketing materials to life.

Biometrics

From fingerprints to facial recognition to age estimation and beyond. For example, if your business sells age restricted products, facial age estimation can automate ID checks to speed up and even automate checkout processes, while also removing stress from team members who find human ID checks daunting.

Anomaly Detection

If your business needs to spot deviations from normal operating conditions then anomaly detection could potentially be valuable. This could apply to vehicle diagnostics, health indicators, software application logs etc or any other scenario where large amounts of data need to be processed to find outlier scenarios. With the introduction of AI there is the potential to identify patterns in data in near real time to attempt to avoid negative outcomes as opposed to using data to support retrospective root cause analysis.

Once you have established that one or more of these technologies could solve a high priority problem and/or opportunity, you need to start thinking about whether your business is ready to adopt AI technologies.

AI Adoption Readiness Assessment

An AI adoption readiness assessment should include at least the following items

Data quality - AI systems are trained on data. If your data quality is low your results will be too

Data integration - if you intend to optimise or automate processes then you will almost certainly need to access data in multiple systems. RAG (retrieval augmented generation) is a perfect example of where AI works across multiple systems.

Data Privacy - have your users consented for their data to be used in the AI system you are adopting?

Data governance - how do you know if your data is good enough? Who is responsible for which pieces of data? Who has permission to access which pieces of data?

IP - do you risk giving away any trade secrets by AI enabling aspects of your business such as product design or software engineering?

Change management - do you have the capacity and expertise to implement new systems? Do you have a plan to upskill your people to adopt new tools and processes?

AI Adoption Case Studies

Case Study - BT

BT recently commenced a much publicised AI transformation. Initially there was a lot of media focus on the anticipated job cuts, but more recently commentary has shifted towards the gains the AI adoption is yielding.

For example, they have used the concept of classification AI to match customer service agents to customers, a technique called intelligent pairing. This works by understanding the context of a customer's need and then matching that customer to a customer service agent who has experience in fulfilling that need. The result should be happier customers because they get a better response more quickly, and a more efficient business because agents spend less time solving each issue.

BT use the concept of content and communication personalisation for a technique called propensity modeling. By understanding their customers through the footprints they leave in product usage and various touch points, sales teams are able to focus on strong opportunities with guidance on what to say to customers, when to say it and on what channel.

Case Study - Google & Siemens

In this collaboration, Siemens, who produce factory automation tools, and Google, who operate cloud based AI services are collaborating to produce AI enabled factory automation. The types of innovations this collaboration is expected to yield range from computer vision systems that can be used to automate quality assurance processes through to predictive maintenance algorithms.

Wrap Up

SME CEOs can create value by adopting AI with our "5 As" framework:

Agree on a value creation opportunity for the business to prioritise
Analyse whether AI can help realise that opportunity
Assess your company's readiness to implement AI systems and processes
Apply those AI systems and processes
Adopt the new AI systems and processes by enabling people with appropriate training and support

What Next?

If you think your business could benefit from adopting AI technologies but you don't have the appropriate knowledge or capacity in house feel free to get in touch today for a no strings chat. Read more about how we help other SMEs with our fractional CTO service.

Article