<\/p>\n
Half the companies racing to put autonomous AI agents into production have quietly admitted something awkward: the data they already own is the part they trust least.<\/p>\n
That\u2019s the buried headline of\u00a0Informatica\u2019s CDO Insights 2026<\/a> survey of 600 data leaders. Among those already running or planning to use agentic AI, 50% cite data quality and retrieval as their single biggest barrier to production. Another 57% still point to data reliability as the wall between a pilot and a real deployment.<\/p>\n The pilots keep multiplying. What sits underneath them hasn\u2019t moved in a year.<\/p>\n Steven Seah<\/a>, who runs South Asia for\u00a0Informatica from Salesforce<\/a>, has seen this play out often enough to call the ending early. Building the agent, he says, is \u201crelatively straightforward.\u201d The hard part is the data underneath it \u2014 keeping it \u201ccurrent, trustworthy, and explainable.\u201d When it isn\u2019t, you get \u201cinconsistent or inexplicable outputs\u201d: the agent doing the wrong thing with total confidence, and no one can say why.<\/p>\n The trap CDOs keep walking into<\/p>\n Most data leaders think they\u2019ve already handled this. A decade of effort went into it \u2014 connecting systems, piping everything into a lake, and wiring up the warehouse or data lakehouse.<\/p>\n None of that is the same thing, Seah argues. \u201cTrusted context is not the same as connected data,\u201d he says. \u201cYou can have data flowing freely across systems and still have agents that can\u2019t act on it reliably \u2014 because there’s no metadata explaining what it means, no lineage showing where it came from, and no quality signal indicating whether it\u2019s fit for the decision at hand.\u201d The plumbing can be immaculate, and the agent is still guessing at meaning.<\/p>\n His fix is to stop treating governance as a box you tick once the data has already moved. Metadata, lineage, and quality should ride inside the data itself \u2014 \u201cnot sitting in a separate layer someone has to consult after the fact.\u201d Do that, and \u201cagents can act with precision.\u201d<\/p>\n The number that should concern every CDO<\/p>\n The cost of getting it wrong, in Seah\u2019s telling, runs to \u201ca USD3.1 trillion annual cost from fragmented, ungoverned data \u2014 not from a lack of data, but a lack of shared understanding.\u201d<\/p>\n He leans on that last distinction, and CDOs should too. Scarcity was never the issue. \u201cMost enterprises have more data than they\u2019ve ever had,\u201d he says \u2014 the trouble is that \u201cmeaning is scattered across systems, each with its own version of truth.\u201d<\/p>\n The symptoms are mundane and costly. In a merger, two teams can\u2019t agree on which org hierarchy survives. In sales, rival reps swear they own the same account. In the supply chain, you can\u2019t trim spend on suppliers you can\u2019t clearly see. “These aren\u2019t edge cases,” Seah says. \u201cThey\u2019re daily operating costs that compound into compliance risk and missed opportunity.\u201d<\/p>\n AI doesn\u2019t soften any of that; it speeds it up. \u201cAn agent can\u2019t act on context that is stale, inconsistent, or locked in a silo,\u201d Seah says. \u201cWhen the foundation isn\u2019t in place, you’re not scaling intelligence \u2014 you\u2019re scaling confusion.” What gets spent down, he adds, is the inventory no company keeps in reserve: “time, money, and trust.\u201d<\/p>\n Headless data management is here<\/p>\nSteven Seah @ Informatica: “When the foundation isn\u2019t in place, you’re not scaling intelligence \u2014 you\u2019re scaling confusion.”<\/p>\n