What is generative business intelligence?

Generative business intelligence, also called "generative BI" or "gen BI," is the practice of applying generative AI to business intelligence processes. Generative BI tools can automate and streamline key data analysis tasks, such as identifying patterns and creating visualizations.

Business intelligence or BI, refers to a set of processes for analyzing business data to inform business decisions. Traditional BI tools and workflows are highly manual, requiring significant time and technical expertise to transform raw data into actionable insights. Stakeholders who lack data science backgrounds are often unable to make full use of BI techniques.

Generative BI allows more people to participate in business analytics. Typically powered by large language models (LLMs), generative BI tools work much like other common generative AI tools, such as ChatGPT or Microsoft Copilot. Users enter natural language instructions and the tool responds accordingly. 

Unlike with traditional BI, users don’t need to learn special programming languages, perform manual calculations or build charts from scratch. They can ask the generative BI tool, in plain language, to conduct advanced analytics and build reports for them.  

In this way, generative BI enables self-service analytics for users across the organization, regardless of skill set. Self-service analytics, in turn, help organizations make more data-driven decisions.

Generative BI is a relatively new technology category. According to one survey, only 3% of organizations report that they have put generative BI into “full operational use.” However, more than half of organizations report they are in various stages of exploring generative BI.¹ Adoption rates are expected to grow as generative BI tools become more refined and more readily available. 

Generative BI and generative AI are not different kinds of technologies or AI models. Rather, one can think of generative BI as a use case for generative AI. Specifically, generative BI is the practice of using generative AI solutions to collect, manage and analyze organizational data to inform business operations.

Generative AI (gen AI) refers to a category of artificial intelligence (AI) and machine learning (ML) models that can create original content—such as text, images or code—in response to a user’s prompt. Generative BI is a type of AI analytics because it applies AI algorithms to process and analyze business data.

Generative BI tools work the same way other generative AI-powered tools do. A user enters a natural language prompt and the tool generates content in response.  

For example, a user might type, “Show me a pie chart with our top 5 best-selling products last year, divided by each product’s percentage of sales.” The generative BI tool would analyze the corresponding data set and return exactly that: a pie chart breaking down top-selling products by percentage of sales.

Most generative BI tools come in one of 3 forms:  

1. General-purpose gen AI models, such as Meta’s Llama, applied to BI tasks. 
    

2. BI platforms with built-in AI models. For example, Amazon QuickSight Q embeds the LLM-powered chatbot Amazon Q into QuickSight, a business intelligence tool from Amazon Web Services (AWS). 
    

3. AI models that are specifically adapted for business intelligence. For example, IBM Project Ripasso is an LLM-powered platform trained on enterprise-relevant content, with built-in data governance capabilities.

While general-purpose gen AI models can perform many BI functions, many organizations opt for the more specialized BI tools and models instead. These typically grant organizations more control over how their data is used.

Features can vary between tools, but common generative BI capabilities include:

• Natural language processing (NLP): Generative BI tools use NLP and chat interfaces so users can submit prompts, query data sets and run calculations without using structured query language (SQL) or other specialized syntax. 

• Custom dashboards, reports and visuals: Most generative BI solutions have authoring tools that allow users to spin up dashboards, data visualizations, written reports and data stories by describing what they need, rather than building them manually. 

• Integrations: Gen BI tools can integrate with various data sources, including databases, data warehouses, data lakes and data lakehouses.

• Recommendations: Many generative BI tools can enrich analyses by recommending related data sets, suggesting related queries, offering feedback on report optimization and providing other guidance. 

• Business glossaries: Some generative BI tools support or integrate with business glossaries. Glossaries allow organizations to define important terms, concepts and processes so that the tool can give responses that are grounded in the business’s unique context.

Generative AI can be used at any stage of the business intelligence process, but it is most commonly used to support data collection, data analysis, data visualization and action planning.

Gen BI tools can help users discover, clean, transform and aggregate data for analysis.  

For example, a user might ask a gen BI tool to put together a report on spending by business unit. The tool would look for relevant data across integrated data sources—including both business-wide financial records and unit-specific records—standardize data point formatting and assemble it all in a coherent report.  

Generative BI tools can consume vast amounts of complex data to surface patterns, answer questions, identify trends and more. This enables users to derive insights from data without performing manual calculations.

For example, the user building a report on business unit spending might ask the gen BI to identify any units that have consistently gone over budget in the last 8 quarters. The user might also ask the gen BI to help identify reasons why these units might be overspending.  

Gen BI can turn the results of its analysis into digestible and shareable graphics and summaries, highlighting key metrics and other vital datapoints and insights.

For example, a user might generate a bar chart that compares business unit spending per quarter against allocated budget to highlight disparities between planned and actual spending.  

Generative BI tools can recommend steps for organizations to take based on data analysis. For example, the tool might recommend breaking down business unit spending on a per-project basis to identify projects that don’t deliver enough return to justify continued investment.

Generative BI tools can enable self-service, advanced data analytics. Users no longer need to master specific programming languages, mathematical formulas or tools to work with data. Instead, they can query, calculate and generate reports through natural language.

Traditionally, business users have relied on data scientists and business analysts to do much of the BI heavy lifting for them. Gen BI removes much of the complexity from business intelligence, allowing users across the business to bring real-world, real-time data into their decision-making. For example:

• Human resources (HR) users can ask generative BI tools to analyze talent trends and make workforce planning recommendations. 

• Finance teams can ask generative BI tools to create more granular forecasting by analyzing revenue at the customer, product and channel levels.

• Supply chain and procurement teams can optimize inventory by asking generative BI to use past trends to predict future buying patterns.

• Marketing teams can use generative BI tools to conduct semantic analysis of customer feedback to derive insights they can use to enhance the customer experience.

• Sales teams can use generative BI tools to analyze the effects of different price points on customer spending. They can use the results to optimize pricing.

Furthermore, the introduction of self-serve analytics frees up data scientists and business analysts to work on more strategic projects. Instead of answering narrow questions that users can now answer themselves, data experts can build new data tools or train proprietary AI models, for example.

Generative BI tools can deliver many benefits, including:

• Improving the adoption of business intelligence tools and practices
• Enhancing business intelligence outcomes
• Addressing data science skills shortages
• Analyzing larger volumes of more complex data
• Reducing the cost of BI efforts

According to one survey, only 25% of users report using business intelligence tools.² Low adoption rates are caused, in part, by the technical complexity of traditional BI processes.

However, generative BI tools allow more users to work directly with their data without having to go through data scientists and analysts. That, in turn, means that more people can use business intelligence to support more data-driven decision-making throughout the organization.

In addition to encouraging more use of business intelligence, generative BI can also enhance the outcomes of business analytics efforts.

Because it can crunch more data faster than a human user or traditional BI tool could, an AI-driven BI tool can often spot trends people might otherwise miss.

Many gen BI tools also prompt users with suggested questions, data and insights to help improve their analyses. And gen BI tools can transform the results of data analysis into visuals and reports for easy sharing and consumption. 

Traditional BI requires a certain amount of data expertise that not everyone has. It can be hard to find enough skilled data scientists and business analysts to fully staff all BI projects.

By enabling self-serve analytics, generative BI tools can help organizations mitigate the impact of data science skills shortages on their BI efforts.

Generative BI tools process larger volumes of data than a data scientist or business user could manually.

They can also process unstructured data, such as documents and images, which makes up an increasing portion of business data. Traditional, rule-based AI algorithms can struggle with data that doesn’t follow a rigid format, but generative AI tools do not have this limitation.

Generative BI can help organizations save time and money by automating many of the most time- and resource-intensive parts of business intelligence, such as running calculations and creating reports. That means organizations can spend less money and labor power on business analytics without sacrificing actionable insights.

While generative BI can deliver many benefits, implementing gen BI tools is not without its challenges. Some of the most common obstacles include:

• Transparency and explainability
• Data security and privacy
• Hallucinations
• Ineffective data architectures

Some generative AI models behave like black boxes, giving little insight into the process behind their outputs. This can be problematic in business intelligence efforts, where users need to understand how data was analyzed to trust the conclusions of a generative BI tool.

Moreover, some regulations, such as the EU AI Act, require that organizations be transparent about how their AI tools process people’s data.

Using generative BI tools that explain their "reasoning," including the data they use and how they arrive at their conclusions, can help organizations maintain transparency and explainability. 

Organizations have both legal and business reasons for prioritizing data security and data privacy. Certain laws, such as the EU General Data Protection Regulation (GDPR), restrict how businesses can use different kinds of data. Moreover, data breaches cost organizations an average of USD 4.88 million per breach, according to IBM’s Cost of a Data Breach Report.

Some generative AI models lack strong data privacy and security measures. Of particular concern is the fact that organizations might not be able to control how these models use their data after they have consumed it.

Generative BI tools with built-in data security and data governance capabilities can help organizations maintain control over their data and prevent unauthorized access.

Generative AI models can experience hallucination. That is, they can make up things and generate false outputs. Hallucinations can potentially derail business intelligence projects, leading to business strategies and action steps that are based on incorrect information.

Organizations can mitigate hallucinations by training generative BI tools on only high-quality, business-relevant data sets. They can also explore other techniques, such as retrieval augmented generation (RAG), which enables an LLM to ground its responses in a factual, external knowledge source.

Like any generative AI model, generative BI tools need access to vast quantities of quality data. A fragmented enterprise data architecture, in which data is siloed off and scattered throughout the organization, can stop a gen BI tool from accessing the data it needs.

An effective data architecture, with the appropriate data storage systems connected in an integrated data fabric, can help ensure that gen BI tools have the data they need to produce quality outputs.