Intelligence Gets Cheap
Michael Homerick@homerick
+4.3%
since Sep 16
The idea
I believe AI will keep taking over the work we pay experts to do, so knowing the answer gets less valuable while judgment, catching failures, and standing behind an outcome get more valuable. People don't disappear, they shift into directing systems and handling the edge cases, which still takes real expertise. I own it across five layers: platforms for directing and checking AI, verification and reliability tools, businesses where an accountable expert signs off, companies that train and credibly verify these skills, and firms making AI processes reliable inside organizations, no single name…
What it owns
- Software where AI agents run tasks inside business workflows and humans verify the results before they take effect. · MicrosoftIt owns the screens where most office workers will direct and check AI, from Copilot in Office to Azure tooling, holding value as long as companies pay for the human steering layer., ServiceNowIt provides the workflow layer where people see AI-driven processes end to end and route unusual cases to humans, a direct beneficiary if reliable execution becomes more valuable than raw capability., SalesforceIt is building Agentforce as the deployment platform where companies run AI agents with humans still in charge, needing enterprises to want that oversight layer rather than running agents without controls., Palantir TechnologiesIts pitch is putting capable AI under human direction inside serious organizations, which aligns directly with the shift from raw intelligence to accountable oversight and controlled deployment., SAP SEIt runs core processes at large companies, so when AI wires into those processes, the reliability checks live in its systems, holding value if it stays the backbone AI has to plug into.33%
- Tools that catch failures in AI-driven work before a customer or patient ever sees them. · DatadogIt lets engineering teams watch whether their AI systems actually work in production, benefiting as more AI in operation drives spending on monitoring and oversight., DynatraceIt automatically spots when complex systems break before customers do, staying valuable as long as large companies treat automated failure detection as required plumbing for AI-heavy operations., CrowdStrikeIt catches bad behavior in systems and increasingly in AI agents, positioning as a first-line spend when companies need security and trust as AI handles more consequential work., GitLabIt provides the review and pipeline rails that keep humans accountable for code as AI writes more of it, depending on companies wanting stronger checks rather than letting AI verify itself entirely., UiPathIt is moving from simple bots to agentic automation where humans set guardrails and handle exceptions, requiring its agentic products to win deals as the shift from pure automation takes hold.22%
- Businesses where a human expert still takes responsibility for the AI's answer, which is where value shifts when answers get cheap. · IntuitIt pairs AI with a live accountable expert through TurboTax Live and QuickBooks, someone who signs off on the answer and owns the outcome., Thomson ReutersIt sells AI tools to lawyers and accountants in professions where a named person takes responsibility for the answer, testing whether AI makes those professionals more valuable per hour., RELXIt owns verified data and analytics that professionals rely on when they cannot afford to be wrong, with value rising as cheap intelligence increases the premium on trustworthy inputs., Verisk AnalyticsIt is the verification layer for insurance decisions where wrong answers cost real money and someone has to stand behind them, working as long as insurers pay for independent checks on AI-assisted decisions.20%
- Companies that teach people to run AI processes and credibly prove they can, as demonstrated ability starts to beat degrees. · CourseraIt is where workers retrain for direct-and-check roles with certificates tied to demonstrated skills, depending on enrollment and enterprise reskilling deals to keep growing., PearsonIt can credibly verify what someone can actually do at scale, positioned to grow as employers shift toward skills-based hiring and demonstrated ability matters more than traditional degrees.6%
- Firms paid to make entire AI-driven processes reliable and governed inside big organizations. · AccentureIt makes whole AI processes reliable for large organizations, with AI bookings providing a direct read on whether companies are actually spending to turn capable AI into work they can trust.4%
- Cash15%