GenAI and Business Intelligence: How we can design decision-making systems and trust them

However, there are also challenges: traditional business intelligence demands deterministic reliability and transparency, whilst GenAI operates probabilistically. So how do we realise the promised benefits of GenAI for our business-critical processes?

The answer is straightforward: if intelligent AI assistants are to orchestrate decisions and optimise processes in future, the technology itself will become secondary once it reaches a certain level of maturity. The decisive factor for the future is trust. Find out here how to bridge the gap between hard data logic and GenAI and reliably integrate smart BI assistants into your business.

“Technological maturity defines what AI can contribute – but it is trust alone that determines what it is allowed to control within our organisations. Systematically built trust is the foundation of the next generation of BI.”

Dr. Philipp Baaden, Customer Success Manager, Comma Soft AG

The Evolution of BI: From Reporting Tool to Intelligent Assistant

BI is a central nerve centre of modern corporate management – whether it is a matter of quantifying market trends, speeding up approval processes, monitoring cash flows or underpinning strategic decisions with data.

When assessing the potential of GenAI for our decision-making processes, it helps to look beyond the current hype. What we are actually seeing is a clear evolution in the way we use GenAI and the responsibilities we entrust to it. To contextualise GenAI within the BI landscape, it is helpful to consider this development in terms of two stages:

• Optimising the status quo: Current developments often focus on making existing systems more intuitive and efficient (e.g. topics such as conversational BI play a role here). Direct dialogue with our data using natural language democratises access, speeds up ad hoc analyses for business units and enables the (partially) automated generation of reports and analyses.

• The far-reaching transformation: In the next step, the focus shifts from reactive analysis to agentic use (e.g. developments centred on the buzzwords GenBI and Agentic BI). This involves AI systems that proactively orchestrate analytical steps, independently recognise patterns, monitor KPIs and integrate themselves as autonomous actors into our value-adding processes.

The potential of these developments to enable new possibilities is immense. GenAI is driving the democratisation of data on a massive scale; even departments without in-depth SQL knowledge are gaining direct access to complex analytics functions. GenAI acts as a constant sparring partner in this process – a virtual expert in data, methods and subject matter, available 24/7. Furthermore, GenAI improves the integration of quantitative data with qualitative and unstructured data (such as market reports, customer feedback, contracts or even social media), which significantly broadens this dimension of insight generation.

When we embed this intelligence into our processes – for example, in financial control or automated marketing campaigns – BI is no longer the analytical bottleneck, but rather the true enabler and accelerator of decision-making systems. A BI agent then works seamlessly hand in hand with a campaign agent or financial control agent and carries out the next steps directly.

“When we talk about assistants in the context of GenAI, there is a clear reason for this: we must not allow responsibility to be diffused. The AI may do the work – but it does not bear responsibility for it.”

The deterministic dilemma and why GenAI requires trust

Amidst all these visionary possibilities (which go far beyond what has been outlined above), we must not overlook a fundamental technological conflict: BI is a deterministic system. When we run an SQL query or calculate a KPI, we always expect exactly the same result for the same data.

We must not naively assume that we can simply unleash a probabilistic system such as GenAI – which generates content based on probabilities – on our data warehouses. A hallucination in a marketing text is annoying; a hallucination in an executive dashboard for liquidity planning is business-critical. Integration is therefore a non-trivial challenge that places high demands on the entire BI process: from data collection through processing to the final analysis.

Added to this is the question: Can we really entrust our data, our core knowledge and our trade secrets to GenAI?

Would you hire a new employee and immediately grant them unlimited access to sensitive strategic data if you were unsure of their values, loyalty and integrity? Hardly.

Our answer to this is clear: only generative AI that is trusted can truly transform. It requires integrated trust layers and the use of autonomous yet responsible trust agents to guarantee security, compliance and transparency at an enterprise level.

Building the bridge: Semantic Layer, BI as Code and putting people at the centre

To bridge the gap between determinism and probabilism and realise the full potential of the combination of GenAI and BI, specific technological and methodological approaches are emerging:

• The Semantic Layer: AI needs context. A semantic model makes implicit (specialist) knowledge explicit. It translates cryptic database logic into business terminology and clearly defined key performance indicators. Only when the AI ‘understands’ what return on sales means in the specific context of your business can it provide reliable answers.

• BI as Code: Is your current BI stack even machine-readable? If dashboards and logic are cobbled together in isolation within proprietary tools, a GenAI has virtually no access to them – its usefulness is limited by this fact alone. Through versioning and code-based BI infrastructures, we create the conditions that enable agents to read, understand and further develop analytical assets.

“Without a semantic model, GenAI is like a brilliant analyst on their first day at work, to whom nobody has explained how the company calculates its KPIs. Before GenAI can provide answers, we need to teach it our business language. The future of BI isn’t just a matter of clicking a few buttons; it’s orchestrated in a way that machines can understand.”

Will we actually still need dashboards in the future?

This question requires a nuanced analysis and serves as a proxy discussion on the symbiosis and benefits of GenAI and our business models and processes. It would be naïve to assume that GenAI limits the scope of possibilities for ‘experiencing data’. Dashboards continue to fulfil the essential purposes of standardisation, management communication and collaboration. They are not an end in themselves, which is why we will see an evolution towards hybrid forms: dashboards will remain, but their utility and accessibility will be massively enhanced by GenAI layers. Whether in terms of design, logic, structure, understanding or interpretation, GenAI will make valuable contributions to existing dashboard logic in many areas.

However, in areas where dashboards are already redundant today and other forms – such as ad hoc analyses – would be required, natural language dialogue systems will dominate; the challenge will be to enable this whilst ensuring maximum transparency.

At the heart of all these considerations, however, one constant remains: us as human beings.

From vision to implementation

So how do we translate these concepts into tangible and reliable value creation? A successful transformation in BI requires a structured approach:

• Identifying quick wins: Where is it already possible to implement or integrate GenAI within your organisation today, and where is the potential impact particularly significant? We specifically seek out use cases that tangibly improve the status quo. In this way, we create initial, positive touchpoints for staff and quickly generate measurable business value.

• Laying the technological foundations: Is your BI stack GenAI-ready? With our GenAI Readiness Check, we assess your data, architecture and processes for potential and identify the blind spots that we need to work on together before scaling up.

• Leadership Commitment & Change Management: We help you develop a medium- to long-term vision and adapt it flexibly to technological leaps. GenAI in BI is ultimately a building block within the overall structure of your organisation. Leaders must take the lead to allay fears and position the organisation for the future.
• Trust by Design: Autonomy requires guidelines. At Comma Soft AG, we provide you with comprehensive advice covering technology, governance and culture. We implement models that are secure, traceable and compliant from the ground up.