Our Take
Deloitte is shipping AI agents into production audit work, but the announcement omits what these agents actually do, how they reduce manual hours, or how they compare to existing automation in the audit stack.
Why it matters
Audit firms process vast document volumes under time pressure. If agentic systems can reduce rework or accelerate evidence collection, margins improve. But 'connected agentic intelligence' is marketing language without measurable claims, so practitioners have no way to evaluate the real productivity win.
Do this week
Audit leaders: Request a technical brief from your Deloitte account team before renewal, specifying what tasks the agents perform, which audit steps they shorten, and which human reviews remain mandatory.
Deloitte Embeds Agents Into Audit Software
Deloitte announced the integration of connected agentic intelligence into Deloitte Omnia, the firm's audit and assurance platform. The company framed the addition as advancing audit capabilities, but provided no technical specification, benchmark, or customer data to support that claim (per press release via PR Newswire).
The announcement name-checks "connected agentic intelligence" without defining what that means in practice. Omnia users will encounter agents, but the press release does not disclose which audit tasks these agents handle, how they route information between systems, whether they reduce manual review cycles, or how they differ from existing workflow automation.
Big Four Audit Workflows Are Finally Adopting Agents
Audit is a natural use case for agentic systems. Auditors spend hours gathering evidence across multiple data sources, verifying compliance, cross-referencing transactions, and documenting findings. An agent that reduces rework, improves evidence traceability, or speeds up preliminary testing could move the needle on per-engagement cost or partner billable utilization.
The catch is that Deloitte has not disclosed whether Omnia's agents actually move either needle. The firm has not published timelines, error rates, human-in-the-loop requirements, or customer deployment data. Without independent verification or customer testimony, the claim remains aspirational.
For Deloitte's competitors and clients, the question is whether this is a feature (agents handle low-risk routine tasks under oversight) or a competitive threat (agents materially shrink audit hours). The announcement does not say.
Audit Leaders Need Specifics Before You Migrate
If you use Omnia or are considering it, ask your Deloitte relationship manager for details: Which audit procedures do the agents execute? What is the human review gate? What is the failure mode if an agent misclassifies a transaction or misses a control exception? Do you retain the ability to audit the agent's work or is the output black-box? What liability sits with you if the agent's output is material to your opinion?
Press releases announcing "agentic intelligence" do not answer these questions. Procurement and quality standards do. Get both before committing to a new version of your core tool.