October 14, 2026
Free Virtual Conference
2:05 pm
Opening
Dr. Carsten Bange, CEO & Founder | DATA festival & BARC
02:40 pm
Alexander Seeliger, Analyst Data & Analytics | BARC
3:10 pm
Instrumenting Business Strategy with Data Excellence: Building a Group-Wide Data Excellence Approach
Anneke Xanke, Head of Reporting & Data Management | ProCredit
How can a business challenge become the catalyst for a Data Excellence transformation? This session shares ProCredit Group’s journey from a business use case to a Group-wide Data Excellence Framework.
We will share how business strategy of balance sheet transformation was focused on customer profitability, and further translated into data requirements, leading to the establishment of a Data Excellence operating model, governance framework, and supporting technology landscape.
The session concludes with practical insights and lessons learned from implementing Data Excellence across an international banking group.
sponsored by
3:40 pm
AI-Readiness: When Your Data’s Next User Isn’t Human
Shree Neve, VP of Operations, North America | ClicData
Business Intelligence was designed around a simple assumption:
humans are theprimary consumers of data. Analysts write queries, dashboards present metrics,and users interpret the results.
AI agents change that model.
As organizations move from dashboards and copilots toward agentic analytics, AIsystems are becoming direct consumers of enterprise data. The challenge is nolonger just how to connect an LLM to a database, but how to give AI the right
context, semantics, permissions, and analytical boundaries to work reliably.
This session will explore how Model Context Protocol (MCP) can act as a structured layer between AI agents and enterprise analytics platforms, enablingagents to discover available data, access governed analytical capabilities,interpret business context, and perform increasingly complex analytical tasks.
The focus will be on the transition from traditional BI to agentic analytics:
BI → Natural Language Analytics → AI Agents → Governed Analytical Actions
Using real customer problems and practical use cases, the session will examine scenarios such as:
- Investigating why a KPI changed, rather than simply reporting the number
- Reasoning across multiple systems such as CRM, ERP, marketing, support,and operational data
- Allowing agents to work with semantic models rather than raw database schemas
- Propagating user identity, permissions, and security controls through AI-driven interactions
- Determining what should actually be exposed through MCP: datasets, metrics, semantic models, tools, or actions
- Moving from “answer this question” toward “investigate this problem” and eventually “take an approved action”
ClicData will be used as a practical reference architecture, including dataconnectors, DataFlows, semantic modelling, AI-assisted exploration, AIDashboard Builder, and MCP.
The session will not focus on MCP as a protocol specification or on AI hype. Instead, it will examine the architectural decisions and real-world challenges that emerge when existing BI environments need to support AI agents.
The central question is:
If AI becomes the next major consumer of your enterprise data, is your BIarchitecture ready for it?
4:10 pm
From Dashboards to Decisions: Making Data Actually Governable
Oana Toropoc, Senior advisor economics | OMV Austria
Organisations have invested heavily in dashboards, reporting, and analytics. Visibility has improved, but decision-making often has not. Data can identify a problem and display it in real time, yet action still depends on authority, incentives, accountability, and governance. This presentation explores what must happen between a data point and an organisational decision, arguing that the last mile of data is not analytics. It is governance.
4:40 pm
From Hours to Seconds: Industrial AI for Defect Prediction in Aerospace Manufacturing
Dr. Aleena Baby, Data Scientist and AI expert | Access e.V.
- A method for turning simulation archives into training data, the most underused asset in most engineering organisations
- A clear view of where domain knowledge belongs in an ML pipeline, and why physics-guided feature engineering can be the difference between algorithms rather than a marginal gain
- Practical patterns for tiered models that degrade gracefully when users arrive with incomplete inputs, and for building ML tools that domain experts actually adopt
5:10 pm
Designing a Production-Grade Multi-Agent LLM Architecture for Structured Data Extraction
Hari Suresh, Principal Data Engineer | eHealth
Why do LLM extraction systems that shine in demos often fail in production? This presentation shows how multi-agent architecture can turn unreliable document extraction into a scalable, production-grade system. We’ll explore specialized extraction, validation, and judge agents, deterministic guardrails, confidence scoring, human review, and observability—bridging the critical gap between an AI prototype and dependable enterprise AI.
*Preliminary schedule, subject to change.
Any questions? Contact us!
