By James Proctor, Co-Founder and Managing Director, The Inteq Group
No. AI agents do not require a consolidated data platform as a prerequisite, because agents consume data in context, at the moment of a specific decision, from the systems where that data already lives. What they require is a process designed around its decisions, including deliberate handling for the cases where data falls short. Platform investment may still be justified, but on its own economics and its own timeline, in parallel rather than in front. The prerequisite framing is the single most expensive belief in enterprise AI, and it deserves a direct examination rather than another year of deference.
What Is the Single Source of Truth, Really?
Here is the observation three decades in this field entitles me to make: the single source of truth is enterprise software's most profitable myth. It has been the justifying vision for the data warehouse, then the enterprise data model, then master data management, then the lake, then the lakehouse, each generation marketed as the arrival point the last one failed to reach. Organizations have spent staggering sums on the journey. I have never met one that arrived, because arrival is not the product. The journey is the product, and everyone selling it is paid by the mile. None of this means the work was worthless; reporting and analytics genuinely improved. It means the destination was never a precondition for anything, and treating it as the gate in front of agent value extends a thirty-year invoice into its fourth decade.
“The single source of truth is a destination the industry is paid to keep moving.”
What Happens When Consolidation Is Structurally Impossible?
The prerequisite belief collapses fastest where consolidation is not merely slow but structurally unreachable. Consider building products distribution, an industry consolidating through acquisition. A serial acquirer is running nine ERPs not because of neglect but because of strategy: the acquisition cadence permanently outruns any integration program, and every completed migration is followed by two new closings. Told that meaningful AI must wait for a unified platform, that organization has been told to wait forever. Designed around decisions instead, the picture changes entirely. Order acceptance, inventory allocation across branches, and customer pricing are decisions, and agents can assemble the context for those decisions across nine ERPs the way the company's best regional managers already do by phone and spreadsheet. The heterogeneity does not disappear. It stops being the gate.
What Is the Actual Prerequisite?
A designed process: the decisions identified, the information each decision genuinely requires mapped to where it lives, and the handling specified for cases where the data cannot support confident action. That design work is measured in weeks, it converts fragmented data from an excuse into an input, and it produces something no platform program ever has: operational value this quarter, from the systems you already own. The honest sequencing question is never platform or no platform. It is which investment must come first, and the answer is the one that is cheap, fast, and reversible.
None of this puts the process side at war with the data organization, and executives should refuse to let it be framed that way. Design-first sequencing is the best thing that ever happened to a data roadmap, because agent-enabled processes generate what data programs chronically lack: a ranked, evidence-based statement of which data, for which decisions, in which processes, actually carries business consequence. The data team stops defending a generalized quality mission and starts executing a prioritized one. The platform, where one is warranted, gets built against demonstrated demand rather than projected virtue, and that is a far easier program to defend in every budget cycle it will ever face.
Mapping decisions to their true data requirements across fragmented systems is a core discipline in our agentic AI consulting services, and it is the analysis that converts the platform conversation from theology into economics.
