Editorial
AI is touching every element of enterprise initiative and the intranet is no exception. One forward-thinking organization told us that AI could make their intranet obsolete. The company is providing information to employees through a simple unified chatbot interface in an effort to sidestep common intranet challenges (i.e., user interface designs, information taxonomies, poor search, channel juggling, etc.).
Ahead of our third SharePoint Online Intranet benchmarking study, we decided to take the temperature on this topic with a selection of our benchmarking partners.
The State of AI and the Intranet?
We asked about their use of AI in any aspect of their intranet. The early results from the first 12 respondents, drawn from a variety of industry contexts, found some common patterns emerging:
- AI depends on content to generate reliable outputs (Garbage In, Garbage Out). Curated, accurate and well-governed information sources — the kind that intranets provide — will become even more critical to successful AI adoption.
- Security and privacy concerns limit the use of AI. The leakage of enterprise data to external AI learning models is of particular concern, which has led to the implementation of in-house AI solutions to provide an approved version. However, for some, especially in government agencies, the use of AI is banned.
- The increased accessibility provided by AI chatbot interfaces, including multi-lingual access, is recognized as a key benefit. Source information must be assured for accuracy and currency.
- Assistance with content creation for “non-professional” writers is also acknowledged as a benefit. It will not replace human editorial judgement and creativity though.
- Overall, AI is seen as an enhancement, rather than a replacement for the intranet. With this in mind, many of our respondents are actively trialling AI solutions, with the aim to provide even more accessible intranet service.
Eating the Intranet, One Chatbot at a Time
Early generative AI adopters have learned that feeding your whole enterprise information repository to the AI beast doesn't work. Reports of AI surfacing unintended content through their chatbot interfaces have become common among these early adopters. The approach is therefore inadvisable, if not downright dangerous!
The emerging alternative is to deploy multiple chatbots that sit across their respective managed information repositories rather than a one-size-fits-all approach.
The formal term for this approach is retrieval augmented generation (RAG) use of the generative AI large language models (LLM). The RAG approach first searches the nominated information repository, e.g. intranet pages, feeding these results into the LLM model to generate chatbot responses.
Intranet site owners can develop chatbot interfaces for their own pages. The scope can be limited to only information contained within these pages. In essence, AI control is achieved through managed exposure to selected content, such as customer service, IT support, HR, Finance or Sales.
Problematic intranet pages are inevitable (e.g. ones with out of date or incomplete content, missing or broken links, poor usability etc.). The best option there is to omit the problem pages from AI support — or at least until their content can be improved and qualified as AI ready.
What Does a Specialized Chatbot Look Like?
We at SWOOP Analytics have developed a number of “mini-bots” in support of both current and prospective customers.
'Security and Technical' Chatbot
One area of long-term aggravation has been “Security and Technical” assessments. Like all providers of cloud-based services, prospective clients are rightly concerned about the protection of their enterprise data. The standard practice is for their IT service to create and provide us with their “security and technical” questionnaires, often containing hundreds of questions. In reality, there really is no industry standard. So-called “standard” questions are regularly expressed in different ways. The end result was a long and drawn out process, for both sides.
To help accelerate the process, we developed a managed repository of required documentation, accumulated over several years. We call it our Trust Center knowledge base. We then attached a generative AI chatbot to the repository to help complete questionnaires faster.
This is an example of how AI can streamline what has historically been a long, tedious process for both clients and providers. The chatbot is currently restricted for internal use. We currently still prefer to have a human in the loop.
Intranet Health Chatbots
Intranet analytics are designed to help internal communications staff better manage their intranets. Like website analytics, the reporting can often be overwhelming, with technology now able to report on every mouse click.
When faced with a plethora of charts and reports, it's natural to ask “what does this all mean?” and “where should I be concentrating my attention?” How can AI help here? The recent ClearBox intranet product review report writes “We like the concept of AI-driven analysis to give greater depth to analytics.”
The above graphic shows our intranet health score report. Selecting the “sparkle” icon launches a conversation related to the report. The underlying “managed information repository” for this mini-chatbot are the support articles related to this report.
Ensure Your Intranet Stays Relevant
Whether you like it or not, AI is coming for your intranet, and poorly managed intranets are quickly becoming irrelevant. Here are some simple steps to stay relevant:
- Ensure intranet site owners understand how AI can help them create high-quality, engaging content.
- Review your intranet sites to identify the well-managed ones. Consider providing a generative AI chatbot front-end to this content.
- Review your analytics for sites that have questionable quality, yet still attract a strong audience.
- Clean up! Look to transform these sites to the quality required for a confident addition of chatbot functionalities. Clean-up includes deleting or updating out-of-date content but also ensuring that everything else has sufficient depth so AI will generate relevant answers.
- Set the aspiration of replacing intranet search with two-way chat engagement, one site at a time, until your whole intranet is “AI enabled.”
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
Editor's Note: Read more on intranet and digital employee experience trends below:
- Artificial Intelligence Trends in the Intranet and Employee Experience Platform Space — Just because an intranet has an AI feature doesn’t mean it’s right for you. Consider how and where you want to use AI, then vet the platforms.
- How to Move Your Intranet From Good to Great — There are plenty of bad intranets and a fair number of good ones. But great ones remain few and far between. Here's what it takes to achieve greatness.
- The Future of Intranets in 2025 — An intranet isn't just a technology project. It’s a living digital ecosystem, powered by people, process and technology. And that won't change any time soon.
Learn how you can join our contributor community.
