IBM, AWS veteran says 90% of your employees are stuck in first gear with AI, just asking it to ‘write their mean email in a slightly more polite way’

Companies are investing heavily in AI but, according to Allie K. Miller, most employees remain trapped in a rudimentary “microtasker” use case—she estimated 90% of users—and studies show 80% of employees use AI at work while fewer than half receive proper training, meaning many annual subscriptions are underutilized. Miller urged a shift up the maturity curve to “Companion,” “Delegate” and especially “AI as a Teammate,” advocating a Minimum Viable Autonomy approach with agent protocols (always do / ask first / never do) and a recommended risk allocation of 70% low-risk, 20% complex cross‑department and 10% strategic tasks to safely operationalize autonomous agents. She warned that within months AI will run multi‑hour autonomous workflows and enable vast numbers of market simulations, implying that firms which fail to embed AI into core systems and governance will cede productivity and competitive advantage over the next decade.

Analysis

Allie K. Miller, CEO of Open Machine, argued at Fortune Brainstorm AI that corporate AI spending is largely wasted because most employees remain in a rudimentary "microtasker" mode; she estimated 90% of users are stuck there, while a Cornerstone OnDemand study found 80% of employees use AI at work but fewer than half have received proper training. Miller's critique ties directly to subscription underuse—annual ChatGPT/Gemini subscriptions become ineffective if users only apply AI for trivial tasks such as rewriting emails.

Miller outlined a maturity path—Companion, Delegate, and especially "AI as a Teammate"—and proposed operational frameworks to scale autonomy, including Minimum Viable Autonomy and agent protocols ("always do," "please ask first," "never do"). She recommended a risk-weighted portfolio for agent tasks (70% low-risk, 20% complex cross-department, 10% strategic), and cited concrete engineering examples such as integrating Codex into Slack to treat agents as coworkers.

She forecasted near-term capability inflection: multi-hour autonomous workflows within months and vastly increased simulation volume as costs fall, implying firms that embed AI into core systems and governance will capture system-level productivity gains. Market signals are moderately positive for the theme, with per-ticker sentiment mildly favorable for IBM (0.3) and AMZN (0.2) and neutral for Google (0.0), highlighting differentiation among vendors and the gating role of training and governance.

Key Decisions for Investors

• Prioritize exposure to enterprise vendors and integrators demonstrating system-level AI deployment and governance (e.g., firms with clear MVA pilots and agent protocols),
• Consider modest overweight to vendors with positive sentiment and enterprise footprints (IBM and AMZN per the signals) while monitoring Google for concrete product integration evidence,
• Monitor corporate KPIs before adding positions: percentage of employees using AI beyond microtasking, breadth of training programs, existence of agent protocols, and early autonomous runtime metrics; reduce exposure to firms that treat AI as a standalone subscription without integration