Take Microsoft’s advancement of Copilot. In under a year, we’ve gone from the excitement of ChatGPT, to Gen AI being integrated into our Office 365 workflow – fitting seamlessly and naturally into our day-to-day.
The majority of generative AI tools that were popular 12 months ago, with a high price-tag, have already been integrated and vastly improved on, with this leading to exponential productivity gains for companies in the long-term.
It’s therefore more important than ever to get your company AI-ready in the short-term: creating a comprehensive data strategy fit for the AI era, considering where to invest time and resources, and being aware of the key mistakes to avoid in the process.
Understanding the landscape
Before diving into the details, there are a few important things to know about the current AI and data landscape – namely, that is it’s evolving all the time.
Vector databases, for instance, are changing the game when it comes to leveraging company data.
Imagine you're at a work event with different groups discussing various projects and interests. A vector database is like a pair of glasses. When you wear them, you can see colourful bubbles representing the topics everyone is discussing, floating above their heads.
These bubbles are vectors, capturing each person's professional interests and attributes. If you're passionate about green technology, instead of wandering aimlessly from group to group, your glasses instantly highlight who shares your interest, by matching your green tech "vector" with theirs. The vector database glasses enable you to swiftly find and join the right conversation, effortlessly connecting you with like-minded colleagues without the need to sift through every discussion.
Vector databases will materially change how datasets are connected. This is a game changer for analysis, but also presents a data governance risk – which is currently the biggest challenge for companies to navigate.
In order to deliver these productivity gains safely, companies must develop proper data governance. Generative AI tools can’t be allowed unmanaged access to any sensitive data, particularly when there is still such opaqueness around how the AI is using or processing it.
The good news, is that this challenge can be relatively easy to solve with the right approach.
How to begin getting your company AI-ready
With so much information to sift through, it can be difficult to know where to start.
Locking in an internal review is a great way to kick off the process, asking yourselves: What is the information/data that is really going to help us maximise our use of generative AI? Where can we have the biggest productivity gains?
Once you’ve decided where you want to focus your AI efforts, the first step is conducting a data audit; tagging all your information to understand the security clearance of each piece of data, and it’s the bedrock of good data governance.
We suggest working from the ground up to isolate which data you need and why, understand where this data sits, and who has access to it. While this may not initially sound like a particularly exciting endeavour, the rewards are well worth the effort.
It’s also important for companies to map out any data that could become an asset. Historical information has gone from being a collection of useless old files, to being data that can offer rich insights drawn from the last few decades.
Invest in getting your core IT infrastructure AI-ready
There is so much excitement around what AI can do, but future gains from generative AI all depend on getting the right infrastructure in place now – not in five years’ time.
In time, everything will be driven by your login and corporate profile, which means that all your data needs to be readily available in the Cloud.
This is reliant on your core, internal infrastructure.
AI-readiness needs to be built in from the ground up and fit into your existing infrastructure. Otherwise, companies risk putting new tech over a “black box” stack, resulting in the business having no idea what the generative AI technology is doing, or what data it is using.
It is therefore prudent to not to invest in generative AI tools outside of your general tech stack, unless there is a really strong or well-thought-through use case.
To create the right internal infrastructure, IT teams should liaise with their managed service provider (MSP) to work out an approach. If you don’t have the expertise in-house, consider hiring a consultant data specialist to help with the project.
Put a clear data policy in place, and embrace using external AI tools like ChatGPT for work with non-sensitive data
It is important to note that there is a difference between utilising external generative AI tools to deliver value-add – something all companies should now be doing – and relying on new tooling in your tech stack. For example, ChatGPT and other generative AI products like Microsoft’s Copilot are easy, “quick win” tools that deliver immediate productivity gains.
The one thing to note here, is that companies need to have an AI policy in place, so everyone knows where they stand with types of data entry into these tools. All policies should include the instruction: “Don’t paste any sensitive company data into GPT”, for example.
Don’t expect getting AI-ready to be a quick and easy process
Thankfully, investing in getting ‘AI-ready’ is not an expensive undertaking, in part because Microsoft and Google want companies to be starting this process. However, it does involve a major time investment.
It may take around six months to get a company’s core infrastructure AI-ready, for example, if its leaning on its MSP to understand how the tech landscape is changing, working out what information is critical, and categorising it properly.
But we advise companies with similar projects to be patient and keep going – the process may be onerous, but it will be worth the investment in the long run. Once you’re AI-ready, your ability to ‘plug and play’ other technology will be maximised, and you’ll be ready to go.
Steve Jones is Head of Data & Analytics atLivingbridge.
