Generative business intelligence, also known as generative BI – which combines generative AI with business intelligence tools – promises to help businesses operate in a smarter, faster way. It empowers non-technical users of BI tools to request information using natural language prompts (NLP) and easily refine the results they receive through further questioning. Although there’s undoubtedly a lot of hype around AI right now, generative BI solves a real problem for businesses. Over the past decade, many have invested considerable sums in BI tools that promise much but often fall short.
Despite the introduction of AI and natural language query features, for instance, crafting a question that will deliver the right results can still be tricky for non-techie types. Key insights may also remain hidden within overwhelming dashboards. Colibri Digital, a UK consultancy at the frontier of AI, big data and cloud computing, works with large and prestigious companies to unlock and harness the hidden potential in their data.
Many users therefore end up downloading data into Excel so they can work with it there.
“That’s very difficult to control and govern, and you end up with data all over the place, which is obviously contrary to the whole data lakes paradigm.” So says James Cross, founder and CEO at Colibri Digital.
Another problem is that simple, recurring requests for data take up a considerable amount of analysts’ time. “It’s very common to have C-level executives sending one-line emails to analysts that say ‘build me a report that shows this’. Often these emails are repetitive every month,” Cross says. “So you’ve got a whole team of analysts – potentially a very large number in a big organisation – whose job is just producing reports for these execs.” A GenAI-powered natural language interface that builds on the data lake and metadata collection work many enterprises have already done could help to address this issue. “Instead of an exec emailing an analyst team, they can open a chatbot and say ‘build me a report that shows X’,” says Cross.
The chatbot will know where to go to get the data if the metadata has already been aggregated and collected. “Moreover, because it’s aware of the history of those requests, and also what other execs are asking for, [it can] make suggestions to enhance that report and enhance the exec’s understanding.”
Cutting through complexity
“What you’re actually trying to do is to spur on curiosity and insight by creating an interface that is as friendly to use as possible,” says Paddy Vishani, director of customer engagement at Colibri Digital.
Such interfaces would undoubtedly benefit users who lack the technical expertise to query data effectively. “What this [GenBI] is doing from a BI perspective is empowering more non-technical users to consume data [that has come from a technical environment], both internally and externally for their own customers,” says James Rush, chief revenue officer at Colibri Digital.
A data catalogue, for example, should be the “yellow pages” of your organisation’s data, says Cross. “But, if the tables are called ‘X_D_J’ or whatever… you’ve got no idea what that means unless you worked on that particular data system.”
Putting a natural language interface over the top of such systems cuts straight through this complexity. “You can say ‘tell me how many beds I’ve got free in this NHS hospital’ and it can answer that question and visualise it [for you],” says Cross.
This means users won’t need to hunt for the particular table or column that holds the information they need. “Anyone can go to a library and see all the books on a topic,” says Vishani. “But it’s much more helpful when a librarian says ‘this is what you’re looking for’. And that’s the key thing we’re now on the cusp of.”
These granular, real-time insights could lead to faster and more effective business decisions. That’s because questioning a GenBI chatbot enables you to quickly get “down to that nitty-gritty detail” and then “get the information into a state where it can be published very quickly”, says Vishani. The result is an organisation where everyone is empowered to make datadriven decisions, leading to faster problem-solving and an enhanced ability to seize opportunities and respond to challenges. A language model trained on an executive’s email history, for example, could also help to unlock highly personalised insights.
“It could say: ‘Well, if you’re interested in the weather price data for this region, perhaps also you’re interested in this correlation, which I found in another region,’” says Cross. “It can make suggestions, but also highlight trends and analyses that perhaps you hadn’t thought of.”
Counting the cost
Training large language models on enterprise data is still expensive, however. “It can cost up to £200,000 for one training run, and you’re not going to get it right first time. So you could end up spending millions,” says Cross. As such, the challenge for the tech industry is to “create something that’s both generic and specific – a ChatGPT that’s optimised for BI use cases, but with a way of tuning it to your business that doesn’t cost a fortune”.
Until then, CFOs and other C-suite executives will need to carefully consider which generative AI use cases are likely to deliver the best ROI. While it’s clear that GenBI and other emerging tools could solve some real business issues, no one wants to get swept up in the AI hype and end up wasting time, effort and financial resources on solutions that don’t deliver transformative results.
“It’s almost too easy to implement it,” says Cross. “You’ve got partners popping up all over the place that claim to know how to do it but don’t. The result is you get something that’s relatively poor quality and not particularly useful, which sullies the whole concept of generative AI, NLP – and, subsequently, generative BI – because of poor implementation.”
Colibri’s affiliation with partners such as AWS enables it to offer customers the most compelling and competitive solutions. That relationship allows for the spark of innovation to run through everything Colibri does. The support and exposure provided to the latest cutting-edge tech means that Colibri’s technology is up to the same trustworthy standards as that of AWS. With the personal service and business insights Colibri offers, plus the trusted support from partners, it offers the innovation of a massive company with the personal touch of a customer-oriented service.
The right partner can make all the difference when it comes to implementing generative AI tools in a cost-effective and results-driven way.
“All major players in the [enterprise IT] market are pitching AI modules as an additional licence,” says Rush. “What Colibri does is empower our customers, defining the use cases for what GenAI to adopt and providing them with a clear path of what the total cost of ownership will look like.”
Crafting the right generative BI tool can transform an organisation’s analytics capabilities as well as its ability to use actionable, understandable data to achieve real business objectives.
Generative business intelligence, also known as generative BI – which combines generative AI with business intelligence tools – promises to help businesses operate in a smarter, faster way. It empowers non-technical users of BI tools to request information using natural language prompts (NLP) and easily refine the results they receive through further questioning. Although there’s undoubtedly a lot of hype around AI right now, generative BI solves a real problem for businesses. Over the past decade, many have invested considerable sums in BI tools that promise much but often fall short.
Despite the introduction of AI and natural language query features, for instance, crafting a question that will deliver the right results can still be tricky for non-techie types. Key insights may also remain hidden within overwhelming dashboards. Colibri Digital, a UK consultancy at the frontier of AI, big data and cloud computing, works with large and prestigious companies to unlock and harness the hidden potential in their data.
Many users therefore end up downloading data into Excel so they can work with it there.
