Pragmatic AI: Useful stuff you can do today

AI is over-hyped - but that’s a little like saying the Internet was overhyped in 1999. While that would have been an accurate statement at the time the Internet did still end up changing everything. The challenge we now face lies in figuring out how to adopt AI usefully, without getting caught up in the sci-fi fever dreams. In this article I’m going to propose some low-investment, high-payoff strategies any organisation can implement with AI right now.

In the use cases explored here all that’s needed is access to one of the major Large Language Models (LMMs), such as OpenAI’s ChatGPT or Anthropic’s Claude, plus a little manual intervention. There are already plenty of companies offering more automated and sophisticated AI-based solutions, and some of them may even be a good fit for your organisation, but “walk before you run” is generally good advice.

There are a couple of important disclaimers though:

1. Take care about providing sensitive data to a commercial LLM. You’ll need to establish clear policies on what data can and cannot be provided to the AI. Data security is one of the primary reasons why many will probably, eventually opt for more bespoke solutions.
2. When they don’t have access to the necessary facts, LLMs will make stuff up (commonly referred to as “hallucination”). This is why you need internal AI specialists to ensure that the AI is prompted correctly and given the information it needs to perform reliably.

The role of AI specialists

Before diving into specific applications, it's worth noting the importance of hiring or internally developing AI specialists within your organisation. These are the people who will figure out how to leverage AI to do practical things in your specific business context. They're crucial for identifying opportunities, ensuring proper implementation, and maximizing the benefits of AI initiatives. How do you find such people? I’d suggest finding current staff that are enthusiastic about AI and asking them to implement one or more of the following use cases. Even if they start with minimal AI knowledge the practical experience will quickly push them up the learning curve. No one really knows how to use AI properly yet anyway.

Pro Tip: If you don’t know how to use AI, at any level from strategic adoption strategies down to, “how do I make it do what I want?”, just ask the AI.  It will be able to give you excellent advice, tailored to your needs.  (See my previous article Consulting is Dead.)

Use case #1: Better meetings

One of the greatest abilities of AI is summarisation. Combined with the fact that we can now easily get excellent automated transcripts from audio, AI can transform the way we conduct and follow up on meetings, offering several key benefits:

• Time Savings: AI-generated meeting summaries can significantly reduce the time spent on post-meeting documentation, allowing team members to focus on action items rather than note-taking.
• Improved Accountability: Clear, AI-identified action items can help ensure that responsibilities are understood and tracked, potentially increasing follow-through on commitments.
• Enhanced Participation: With AI handling the minute-taking, participants can engage more fully in discussions, potentially leading to increased meeting participation and idea generation.
• Objective Assessment: AI analysis of meeting effectiveness against agendas can help teams improve their meeting practices over time, potentially reducing unnecessary meetings.

Use case #2: Accessible corporate knowledge

Excellent organisations have excellent documentation, but sadly this isn’t enough. Far too often, carefully compiled corporate knowledge rests in unopened documents in repositories that were forgotten about months or years ago. AI can change all that by making everything instantly discoverable with no need to navigate SharePoint or deduce keywords. By using AI to make company knowledge more accessible, businesses can realize significant benefits:

• Faster Onboarding: New employees can get up to speed more quickly when they have AI-assisted access to company policies and procedures.
• Reduced Support Burden: With employees able to quickly find answers to common questions, HR and IT departments could see a reduction in routine queries.
• Improved Compliance: Easy access to up-to-date policies can increase policy adherence, reducing risks associated with non-compliance.
• Enhanced Decision Making: When employees can quickly access relevant company information, decision-making processes can be accelerated.

Use case #3: De-risking compliance

Building on the previous use case, AI can be a powerful tool in managing regulatory and standards compliance:

• Risk Reduction: AI-assisted policy reviews can help identify potential compliance issues faster than manual reviews, potentially reducing regulatory risks.
• Cost Savings: By automating initial compliance checks, companies may be able to reduce the time and cost associated with compliance management.
• Adaptability: AI can quickly analyse new regulations and suggest policy updates, allowing businesses to adapt to regulatory changes more rapidly.
• Comprehensive Coverage: AI can process and cross-reference large volumes of regulations and company policies, potentially increasing the thoroughness of compliance checks.

Use case #4: Better customer support without spending a fortune

While most large organisations have probably already implemented sophisticated help desks to assist with customer service, AI can make this sort of functionality available to even the smallest organisations. Once you have given an AI access to your product documentation and support materials, even the part-time, temp receptionist on your front desk will be able to handle the most obscure of queries.  

The beauty of this approach is its immediacy and accessibility. With nothing more than access to a commercial AI tool and your existing product documentation, you can start enhancing your customer service operations. There's no need for complex integrations or specialized AI development – just a thoughtful application of readily available technology.

Of course, if you want to go a smidge further, it’s easy and cheap to embed the same knowledge in a chatbot you can add to your website in a few minutes.

Conclusion

These AI applications offer tangible benefits that can significantly impact a business's efficiency, compliance, and customer satisfaction today (quite literally today if you want). While you're exploring these low-risk, low-cost approaches your emerging AI specialists will be developing the core skills needed to assess how and when to move on to more sophisticated implementations.

Remember, AI is not magic, and anyone that claims to know how best to use it in your specific circumstances is probably being creative with the truth. The businesses that benefit most from AI will probably be those that approach it not as a revolution, but as an evolution in how we work - one that requires thoughtful integration into existing processes and a commitment to ongoing adaptation and improvement.

Note: Posts with images get more views, apparently. I’ve used AI-generated images in the past, which was fun for a while, but one of the things AI is doing is creating a world in which reality and authenticity are becoming luxury goods. Thus I now choose to use my own photos. I’m not claiming they’re relevant or good photos, but they are my photos, taken thoughtfully and with care by an actual person.