Creating AI value with enterprise context

An MIT Technology Review Insights poll found that only 16% of surveyed organizations are extracting transformative value from AI; 40% report moderate gains in specific tasks or teams, 21% meaningful gains, and 23% minimal or no returns. Most respondents are piloting AI (38%) or using it to assist employees (35%), while 12% have AI agents in production and 3% use AI for most work in key processes. Atlan co-founder Prukalpa Sankar identifies a lack of business context as a key obstacle to AI usefulness inside organizations.

Analysis

AI value capture shifts toward integration, not model access. The investable implication is a longer enterprise-software implementation cycle: fragmented data, permissions and workflow definitions can keep agents in pilots while organizations continue paying for copilots and experimentation. That weakens near-term assumptions that AI automatically drives broad seat expansion or autonomous-workflow revenue, but does not by itself invalidate infrastructure demand; experimentation can consume compute before productivity is measurable.

If the bottleneck is enterprise context, value may accrue to vendors controlling systems of record, identity and distribution—such as Microsoft, Salesforce and ServiceNow—because they can bundle connectors, governance and workflow execution. Conversely, standalone context/catalog layers such as Atlan may face both demand friction and platform bundling; services firms may win implementation work, though labor-heavy deployments could dilute customer ROI and slow rollouts. These are conditional economics, not proof of vendor revenue capture.

The evidence is weak as a company-level signal: the survey is self-reported and published by a custom-content arm, and it does not establish budgets, paid usage, renewals or productivity outcomes. Near term (days), little basis for repricing a single name. Over 1–3 months, watch earnings commentary on AI attach, paid adoption and deployment-to-production conversion. Over 6–18 months, the key test is whether governed context enables repeatable workflow automation or remains bespoke integration. Thesis weakens if vendors report accelerating production deployments and measurable customer ROI; it strengthens if pilots persist while AI-related spend or guidance is cut.

Key Decisions for Investors

• Do not trade the survey as a standalone bearish AI signal. Treat it as a reason to discount near-term autonomous-agent revenue assumptions, not a reason to short compute demand.
• Relative-value watch: favor Microsoft, Salesforce and ServiceNow over less differentiated AI application vendors only where valuation does not already price successful monetization. Reassess after the next 1–3 months of earnings: require evidence of paid AI adoption, renewal uplift or production workflow conversion before adding exposure.
• Monitor context-layer economics before backing standalone vendors: verify customer deployment time, retention, paid expansion, integration cost and whether major platforms bundle comparable governance/context features. Bundling or persistently bespoke implementations would falsify the independent-layer upside case.
• Main risk to the cautious view: a rapid rise in production agents with demonstrable labor or cycle-time savings could accelerate enterprise budgets and make today’s pilot-heavy picture stale; sustained pilot status alongside weaker AI guidance would reinforce it.