As I ponder the state of organisational knowledge in the digital age, I often recall a phrase: "data-rich, insight-poor." We amass terabytes of research, yet the signal-to-noise ratio often hinders, rather than helps, decision-making. So I jumped with both feet when it felt at last possible to create an AI-powered insight assistant. Such a proof-of-concept could be a step towards truly democratising knowledge within an enterprise.

At that time, the challenges we faced at loveholidays were familiar to many startups, scale-ups and larger organisations: insights locked within departmental fiefdoms, employees spending valuable time hunting for relevant data, and a reliance on manual processes that stifled agility. Sometimes the symptoms of a larger problem: knowledge hoarding.

The initiative wasn't just about applying the latest technology; it was about orchestrating a set of tools to unlock trapped potential. OpenAI's natural language processing became the engine, ingesting and distilling complex research reports into easily digestible summaries. RunBear acted as the conductor, seamlessly integrating this intelligence with existing data repositories and, crucially, the Slack communication platform.

Why Slack? Because that’s where the conversations were already happening. Embedding the insight assistant directly into the daily flow of communication meant we weren't asking employees to adopt a new tool, but rather, enhancing the tools they already used.

To help with adoption, we gave the new assistant a name: Polus — a mythological entity of knowledge, intellect and farsight.

Within days product managers, designers, marketeers, the heads of departments, were using Polus to fill their knowledge gaps related to our customers, the markets, our competitors, and challenge their own, and each others' assumptions.