Some companies give AI to everyone. This CEO is doing the opposite

Paysend CEO Ben Chisell estimates AI lets the company deliver around 25% more engineering work with the same-sized team, but limits access and requires employees to show concrete value. Other firms report AI-assisted work taking days instead of weeks, while U.K. government research found 75% of AI-using businesses reported improved productivity and 77% reported no revenue change. The article highlights uncertainty over whether AI efficiency gains translate into financial results, alongside examples of companies using added capacity to take on more work rather than cut staff.

Analysis

The investable signal is not “AI adoption,” but whether productivity becomes additional revenue, lower unit costs, or merely more output that requires review and rework. For software and payments businesses, the likely first payoff is capacity to ship more products and serve more customers without matching increases in engineering headcount; near-term margin expansion is less certain if firms reinvest that capacity or retain staff. Poorly governed AI could instead raise QA, security, and incident costs, making output volume a misleading productivity measure.

For Microsoft (MSFT), broad license deployment supports distribution but does not establish durable customer ROI or willingness to pay at renewal. For Lloyds Banking Group (LYG), the key question is whether AI improves cost-to-income or service outcomes, not employee activity. Sage (SGE) may benefit from faster product development, but company productivity claims need confirmation in release cadence, customer adoption, and financial performance. Paysend’s approach is a useful operating hypothesis, not evidence of sector-wide economics.

Over 1–3 months, watch earnings commentary for measurable changes in engineering throughput, hiring, delivery times, and AI-related spend. Over 6–18 months, advantage may accrue to firms that redirect capacity into customer-facing products while maintaining human accountability; indiscriminate automation could create a quality and trust liability. The contrarian point: seat counts and time saved may be overvalued as proof of monetization, while the longer-term benefit—more ambition from the same workforce—may be underappreciated. No high-conviction directional trade follows from this evidence alone.

Key Decisions for Investors

• Keep MSFT on an AI monetization watchlist, but do not treat enterprise license deployment or usage as proof of incremental revenue. Verify paid-seat growth, renewals, and customer ROI in forthcoming disclosures.
• For LYG, monitor whether AI translates into measurable operating leverage or better service outcomes; employee adoption alone is not a thesis catalyst. Reassess if management quantifies sustained cost-to-income improvement.
• For SGE, look for evidence that AI-enabled development improves product release cadence or customer retention without a rise in quality issues or support burden. Treat productivity claims without financial corroboration as non-confirming.
• Avoid a sector-wide labor-displacement trade for now. The thesis would strengthen if companies show durable headcount savings without slower delivery or worse customer outcomes; falsify it if AI-related spend rises while productivity, margins, or service metrics fail to improve.