Intranets – are they in a state of change? Is the way they work going to categorically shift forever? 

Swoop Analytics’ masterclass: ‘Getting your intranet ready for AI’ certainly raised those questions, but let’s start with the practical tips before moving onto the philosophical. 

So, the most basic question is: how do you integrate AI into your intranet? For the purposes of this webinar, we largely centred on the implementation of an AI chatbot or lookup function. 

It was reassuringly established that most intranets have a long way to go before they’re AI ready. In fact, 96% of the attendees identified themselves as still being in the ‘early stages’ of adoption.  

It’s unsurprising really considering the scale, time and investment that’s required is significant. More than that, there’s a strong governance piece too – bringing AI into your intranet means opening up access to confidential, and live organisational information, which necessitates care handling.  

Now, where do you actually start on this intranet journey? You should start by starting small. Build a solid foundation first.  

This leads to…

The golden rule of getting your intranet ready for AI

AI is only as good as the content that sits behind it. 

AI draws from the information it’s fed. Feed it something incorrect? Poorly worded? Unclear? Guess what you’ll get back…  

For writing content for AI structure matters. Use clear headings, subheadings, bullet points, and detailed but straight-forwardly written information. Homogenous paragraphs with no clear outline will yield poor results. 

The Swoop team made a particularly important point here – regardless of the content quality you have, AI always presents its output as fact. This is known as the ‘credibility trap’, and is why it’s essential to get your content right before bringing AI in. An AI chatbot confidently delivering incorrect answers to your people is a real risk, you can imagine the amount of things that may go wrong. 

That’s also why educating your colleagues on writing for AI will be essential. It’s likely that you will have to revisit and rewrite your existing content– so, before doing so, brief the to-be editors to ensure they understand the requirements. Put a stress on clarity, consistency and factual accuracy. This matters across the whole intranet, not just the pages you own. Whether Finance or HR others look after certain sections, everyone needs to be aligned. Inconsistent governance will mean that the quality of AI implementation will suffer. 

This all connects to current challenges we already face with intranets: content quality, duplication, information architecture, permissions and change adoption. Overhauling your intranet in preparation for AI will expose many of these issues, and you can treat it as an opportunity to address them. 

The AI Readiness Index

Swoop also introduced their AI Readiness Index, weighted across three areas accordingly: 

• Content readiness: 60% 

• Engagement readiness: 30% 

• Search effectiveness: 10% 

A score of 54/100 or above, in their eyes, means you are ‘ready for AI’. Interestingly, the average across organisations sits at 51.1, and 54+ puts you in the top 20% globally. Basically, don’t panic if you feel behind, you’re probably not. Don’t be afraid to just have a crack at improving your intranet to get there. 

New challenges

In all this change what are some new intranet challenges that could emerge?  

• Prompt and answer accuracy – significant testing and iteration will be needed.  

• Bot training – it won’t be perfect first time, and as your organisation, intranet and the technology itself evolves, this will become an ongoing, perhaps Sisyphean job 

• Attribution and accountability – this is particularly interesting. When a bot produces an answer, who owns it, who takes responsibility? A new question around culpability that didn’t exist.  

• Ethics – a pretty obvious one but what are the implications of exposing organisational information to AI at this scale and depth? 

• Behaviour analytics – measurement and insight will shift from what people view to the questions they ask. This could potentially unlock richer insight into what really matters to your people.  

So, a lot in motion, and a lot to come, and most of us are still at the beginning of this journey. 

How intranet content will change forever

This brings us to that philosophical question. Should we be writing for humans or for AI? 

This question sparked hot debate. Some said humans only, with one memorable response being “humans, and AI should learn how to adjust to that”, which is an interesting point.  

Others argued AI, after all, if we are going to end up leaning primarily on AI as our source of information, then surely all the foundation material should be written for AI to make it function as well as possible.  As one attendee said: “the intranet will be less a website to read; but a repository for AI interfaces to use.” 

Overall, the consensus was that, ultimately content should be primarily written for humans. After all, as communicators, that’s who we’re trying to reach. However, really, we need to write for both.  

Keeping that authentic human voice, but structuring that for AI? An exciting challenge for the future of intranets.