For expert knowledge owners, knowledge provides a competitive advantage. Yet, this valuable information and insight is typically locked away in forums, behind paywalls, buried in web archives and servers.

With generative AI search, it’s now possible to train your own ‘digital expert’ that allows people to converse with your specialist knowledge. Generative AI enables deeper and more contextual searches, meaning vast bodies of expert knowledge can be interrogated quickly and efficiently at a subject level. 

There is a fundamental change happening in the way people interact with information. We believe this is a game-changer for anyone who owns the top 5%-10% of knowledge in their specialism and will become the default search method within the next two years.

Suppose you want to find a solution for reducing carbon in Smart Cities.

Conventional search will give you links to resources based on keywords, categories and tags. You’ll then need to go through every link to collate the information, review the references and articulate your answer.  It will likely take 4 - 8 hours.

Generative AI search will give you an immediate written answer to your question with the exact references, whether they exist in an article, presentation, discussion thread, or video. 

This is not ChatGPT, which is trained on the world’s knowledge. To create a credible digital expert, the AI is trained on your knowledge and configured to deliver against your specific outcomes, making it focused, smarter and more useful to anyone looking for expert answers.

We recently trained a generative AI Beta for The Digital Twin Hub (DT Hub) on sources relating to digital twins dating back five years, encompassing PDFs, reports, videos, websites, and discussions on a community platform. 

To test its capability, the DT Hub’s human expert asked our digital expert a series of questions, including how it would structure an important government presentation they were giving the following week. 

The AI answered 18 out of 20 questions, and because it had been trained with the DT Hub data, the human expert confirmed that its answers were credible - including the presentation outline, which was described as 100% accurate. 

Testing also highlighted gaps in DT Hub’s content (the missing two answers), which they said will inform future content production. We are continuing to work with DT Hub to refine the model with more data.

“For an early-stage beta, the accuracy of the answers is remarkable,” said James Wylie, Temper's Managing Director. “To be able to train a digital expert that gives answers that the human expert says are credible is going to be a game-changer for knowledge owners in industries where growth is fuelled by information sharing such as innovation & science.”

What excites us is that a digital expert can be finetuned to guide users through a more immersive and valuable knowledge discovery journey that also delivers against strategic commercial outcomes.

Knowledge owners can also use it to identify knowledge gaps and opportunities to deepen subject authority while protecting the integrity of their knowledge from general AI models.

This approach is so much more than simply better site search and organisations that master it will become the information leaders for their specialism.