By Guest contributors |
Leaders deciding how quickly to deploy AI agents need to know whether their business intelligence (BI) foundation is ready. Microsoft’s own research points to the rise of the frontier firm, an organisational model built around human-agent teams. AI agents are only as reliable as the data they reason over. As organisations look to Microsoft tools including Fabric, Power BI and Copilot to support new ways of working, many are discovering their existing BI estate was never designed for autonomous decision-making.
Walk into a typical large enterprise and you’ll find reporting scattered across Tableau, Qlik or MicroStrategy, with business logic hard coded into hundreds of individual dashboards. A human analyst can navigate that fragmentation. An autonomous agent cannot. Ask an agent which accounts are at risk and it needs one governed definition of what ‘at risk’ actually means, not five conflicting versions living in different teams. Static dashboards were built to be read, not to be reasoned over. That is why agentic readiness is, before anything else, a data architecture problem.
Legacy BI is also expensive to maintain. Enterprises routinely pay three to five times more to license and run legacy BI than a modern platform would cost, before the hidden overhead of separate servers, gateways and specialist skills is even counted. Every dollar spent keeping legacy reporting alive is a dollar not invested in AI. Modernising typically frees 30 to 40 per cent of that spend, which can be redirected into AI and data engineering. Many organisations already own the platform they would migrate to.
When migration becomes unavoidable, the instinct is to lift and shift by cloning the existing dashboards onto a new tool. Recreating yesterday’s fragmentation on tomorrow’s platform delivers none of the modernisation and leaves agents with nothing solid to stand on.
The organisations getting this right treat migration, modernisation and rationalisation as a single exercise. They rationalise the estate first, retiring redundant reports. In one programme, dashboards were consolidated by more than a third. They modernise with Fabric and Power BI. They build a governed semantic layer, one certified source of business definitions and metrics that Copilot in Fabric, natural-language queries and AI agents can all draw on with confidence. A metric is defined once and consumed everywhere, so a dashboard, a natural-language question and an autonomous agent all return the same number. That consistency, backed by clear lineage and governance, is what turns raw data into something an agent can genuinely be trusted to act on.
Migrations succeed or fail on adoption, so the work needs to be run as a change programme that includes retraining analysts, standing up a centre of excellence and shifting the culture from analyst dependency to genuine self-service. Purpose-built accelerators and automated conversion utilities can compress the timeline considerably, resulting in a workforce and data foundation ready to act.
The frontier firm will be built on data agents can trust enough to act upon. Migrating BI creates the foundation for autonomous decision-making while reducing the cost of legacy reporting. Enterprises that recognise this now will be better positioned to scale agentic AI reliably.
Neelabh Saxena (left) is a principal and Satish Jha is an associate principal at ZS
Read more like this, including news, interviews, case studies and insight from industry experts, in the Autumn 2026 issue of Technology Record. To get content like this on a regular basis, subscribe to the print edition or free digital edition of the quarterly Technology Record magazine.