By the Editorial Desk
In-Depth Analysis & Enterprise Technology Reporting
Executive Overview
Consider the legal realities of property management in California. Under state law, a landlord has precisely 21 days following a tenant’s departure to either return the security deposit in full or provide an itemized statement detailing lawful deductions. Fail to meet this deadline in bad faith, and the statutory penalties can mount to twice the original deposit amount.
Crucially, the law remains entirely agnostic regarding the nature of the entity responsible for the delay. It cares not whether the notification letter was meant to be generated by a human property manager or dispatched by an automated software routine. When a deadline is missed, an accountable party must answer for the failure.
This legal and operational reality is a primary preoccupation for Rick Folgmann, Chief Operating Officer at Get Covered, an insurtech company operating at scale across the rental housing sector. Today, Get Covered deploys 16 specialized AI agents running across more than 3 million rental units nationwide. Reflecting on the deployment process, Folgmann notes a telling reality: while engineering the underlying technology presented its own hurdles, the far more complex challenge was determining precisely who each agent reports to.
As businesses race to integrate autonomous software into their core operations, this organizational question is becoming the defining management puzzle of the decade. According to Deloitte’s State of AI in Enterprise survey, roughly 23% of companies currently utilize agentic AI at least moderately, with an overwhelming 74% anticipating adoption within a two-year window.
Yet, this surge in enthusiasm is met with stark warnings from enterprise analysts. Gartner has famously predicted that more than 40% of all agentic AI projects will be outright canceled by the end of 2027. This impending high rate of project failure is not fundamentally driven by technological limitations or model failures. Instead, industry veterans argue it stems from a profound oversight: organizations are deploying autonomous software without establishing the rigorous governance, procedural guardrails, and managerial hierarchies traditionally reserved for human employees.
Detailed Chronology: The Evolution of Workplace Automation
To understand how enterprises arrived at the current crossroads of agentic AI, it is necessary to examine the rapid evolution of workplace automation over the past decade.
Phase 1: Deterministic Scripting and RPA (2015–2020)
For years, enterprise automation was synonymous with rigid scripts and Robotic Process Automation (RPA). These systems operated on strict "if-this-then-that" logic. They were exceptionally good at repetitive, high-volume tasks—such as scraping data from legacy spreadsheets or populating invoice fields—but possessed zero contextual awareness. If a user deviated even slightly from expected inputs, the automation would crash, requiring immediate human intervention to resolve the exception.
Phase 2: Generative AI and the Chatbot Boom (2022–2024)
The public release of foundational Large Language Models (LLMs) in late 2022 fundamentally altered expectations. Suddenly, machines could comprehend unstructured text, draft emails, and converse with human fluency. Enterprises rushed to deploy conversational chatbots and generative text tools. However, these systems were largely passive; they waited for human prompts, generated content, and stepped back. They were tools rather than actors.
Phase 3: The Rise of Agentic AI (2024–Present)
We have now entered the era of agentic AI. Unlike their predecessors, AI agents do not merely suggest solutions or answer prompts—they take autonomous actions. Equipped with specific objectives, access to application programming interfaces (APIs), and the ability to chain thoughts together, these agents can execute multi-step workflows across disparate enterprise systems without continuous human oversight.
An agent can now monitor vendor compliance, evaluate insurance documents, initiate communications, and update customer relationship management (CRM) systems dynamically. Yet, this newfound operational freedom has created a governance vacuum. As software transitions from a passive assistant to an active decision-maker, traditional enterprise hierarchies are being severely tested.
Supporting Context & Metrics: The Promise vs. The Reality
The divergence between enterprise ambitions and execution capability is vividly illustrated by recent data from major research and advisory firms.
The Metrics of Adoption and Risk
- Deloitte (2026): 23% of enterprises currently utilize agentic AI in a moderate-to-heavy capacity, while 74% expect to integrate these systems into their operational workflows within the next 24 months.
- Gartner (2027 Forecast): Over 40% of all agentic AI projects initiated by enterprises are projected to be canceled or abandoned by the end of 2027, primarily due to unclear value propositions, regulatory exposure, and a lack of proper governance frameworks.
- McKinsey & Company: Recent research on the state of AI indicates that the elite minority of companies achieving significant, measurable bottom-line returns from AI implementations—representing roughly 6% of total respondents—are nearly three times as likely to have fundamentally redesigned their underlying business workflows rather than simply layering software onto legacy processes.
The Core Impediment: Process Redesign
According to Folgmann, true operational transformation begins with codifying rules into explicit, unambiguous terms. In property management, tasks such as chasing vendor certificates of insurance (COIs) consume an average of three hours per property every week. While tedious, the task is entirely rule-bound, governed by explicit insurance requirements embedded in vendor contracts.
When companies attempt to automate messy, undefined processes, failure is virtually guaranteed. As Folgmann observes: “If a new hire couldn’t follow your description of a decision on the person’s first day, an agent won’t follow it either.”
Official Statements and Industry Insights
Enterprise leaders are increasingly vocal about the need for structural discipline in AI deployments. The consensus among top operators is that managing software agents requires the same rigor applied to human staff in regulated industries.
On Workflow-Specific Autonomy
One of the most common misconceptions among technology executives is treating "autonomy" as a binary switch—a fixed characteristic of a software product. In practice, mature organizations calibrate autonomy granularly based on the specific risk profile of each workflow.
At Get Covered, agents operate across a tiered spectrum of independence:
- Low-Risk Automation: Routine, non-controversial communications—such as sending an automated reminder to a vendor whose certificate of insurance expires next month—are executed with total autonomy.
- High-Risk Oversight: Complex, sensitive interactions—such as responding to a resident’s formal request for an emotional support animal—are strictly gated, requiring human review and approval before any action is taken.
Leadership teams are advised to evaluate every workflow touched by an AI agent through a simple risk-assessment lens: If this action is executed incorrectly, how long will it take to discover the error, and what will the financial, legal, or reputational cost be by that time? Workflows characterized by rapid error detection and minimal downside earn greater operational freedom; those with high stakes require strict human-in-the-loop controls.
On Individual Accountability and Ownership
In traditional financial services and underwriting, automated credit-scoring models were never permitted to serve as an excuse for a policy failure. Regulators and consumers demanded clear explanations, meaning every automated decision had a human owner behind it.
This principle is even more critical in the age of agentic AI, where software systems operate faster, broader, and across more interconnected databases than legacy underwriting engines ever could.
To prevent systemic drift and operational silos—where multiple departments deploy independent agents that inadvertently contradict one another—every enterprise agent must be assigned a named human owner. This designated manager is responsible for:
- Regularly auditing the agent’s error logs.
- Formally evaluating performance metrics.
- Deciding when to expand, restrict, or suspend the agent’s autonomous privileges.
- Answering directly to internal compliance committees or external regulators when anomalies occur.
Future Outlook: Navigating Toward 2027
As we look ahead to the remainder of the decade, the trajectory of enterprise technology is clear. Foundational AI models will inevitably continue to improve, becoming faster, cheaper, and more multimodal. However, institutional clarity—the ability of an organization to define who decides what, and under whose authority—will not improve automatically.
Gartner’s prediction regarding the widespread cancellation of agentic AI projects serves as a definitive warning flare for corporate leadership. The initiatives that fail will not do so because the underlying algorithms were flawed, but because management failed to define value, control risk, and establish clear lines of accountability.
For executives and decision-makers charting their strategic roadmaps, the path forward requires a shift in mindset. AI agents should no longer be viewed merely as IT projects or software deployments. Instead, they should be integrated into the organization as digital staff members operating within regulated roles.
By writing down explicit rulebooks, limiting operational scope by workflow, and assigning a named human manager to oversee performance and audit outputs, companies can build the evidentiary foundation required to justify expanding autonomous responsibilities over time.
Organizations that embrace this disciplined, management-first approach will successfully harness the transformative power of agentic AI. Those that treat software as a set-it-and-forget-it novelty will spend their time explaining to stakeholders why their digital workforce took actions that no human ever approved.
Rick Folgmann is the Chief Operating Officer at Get Covered, an insurtech innovator delivering streamlined property coverage solutions through advanced digital platforms. He brings over 25 years of executive leadership experience spanning the financial services and insurance sectors.
