[IT DAILY] FT's Warning: AI Makes Apps, But It Fails to Create Value.

Financial Times (FT) chief data journalist John Burn-Murdoch recently pointed out the structural paradox of the current AI boom in his contribution article, ‘How much value is AI actually creating?’.

The data he presented clearly reveals this structural paradox. Since 2025, when the Agentic AI era began in earnest, the number of iOS app launches has almost doubled, reaching about 180% of the 2024 average. This is the result of AI coding tools allowing anyone to create apps quickly. However, during the same period, the number of apps with significant usage did not increase by even one, and the number of app reviews actually decreased.

Burn-Murdoch defines this as the ‘confusion of activity and productivity.’ While it is true that AI speeds up drafting, coding, and summarizing, that is merely an increase in activity; true productivity must be a useful output relative to total inputs, including error correction, review and verification costs, and risks.

his goes beyond simple statistical observation and suggests that the axis of competitive advantage is fundamentally shifting. In a software-native environment, competitive advantage no longer lies in the ability to write code or develop software itself. This is because generative AI has virtually eliminated the barrier to entry for coding. Now, real value is shifting to the operational capability to deeply understand complex and heterogeneously intertwined business systems in the real world and to automate them reliably without errors.

[Bottleneck Exacerbated by Acceleration — Why AI Reveals Organizational System Problems]

Burn-Murdoch attributes the reason why AI fails to create value not to a defect in the technology itself, but to problems in organizations and systems. Increasing throughput using AI proportionally increases the total amount of error and uncertainty. Generative AI's hallucination phenomenon is a prime example. Ultimately, a bottleneck structurally occurs where more human time and cost are input to review and verify the increased output, control quality (QC), and comply with regulations.

There is also a warning at the business model level. The cost of using AI currently felt by companies is kept low thanks to massive subsidies from big tech model providers such as OpenAI and Anthropic. However, a structural risk remains latent that if they begin to charge normal prices to make up for cumulative losses, the cost of utilizing AI may exceed the cost of operating human resources.

This aligns with the J-curve effect of past technological revolutions. When computers and spreadsheets were first introduced in the 1990s, productivity indicators actually stagnated for the first few years. Value became visible only after complementary assets, such as human capacity to handle technology, restructuring of organizational workflows, and innovation in business models, matured.

The current AI craze is also in a state where the organization's acceptance system lags behind the speed of technological development. The FT's diagnosis is that it is difficult to create true value by simply layering AI on top of existing business processes.

The message is clear. The current structure, where trillions of dollars are concentrated only on the AI infrastructure and supply side, is not sustainable, and if it fails to prove usefulness that actual users can recognize, it could be hit directly by the AI bubble theory.

[Enterprise Reality: DB Query Bottleneck is a Common Challenge for All Industries]

Large commercial databases sit at the core of the IT infrastructure of all enterprises running RDBMS-based systems, regardless of finance, public, manufacturing, or distribution. These systems process hundreds of millions of transactions a day, and a response delay of just a few tens of milliseconds can lead directly to customer churn, compliance violations, and service disruptions.

The core cause of DB performance degradation is inefficient SQL queries. Problems such as excessive Buffer Gets, unoptimized execution plans, and unoptimized Indexes quietly accumulate while the system is operating and explode at the threshold.

As AI increases throughput, the query load flowing into the DB also increases, resulting in a paradoxical situation in the field where DB bottlenecks worsen after AI adoption. A DBA's SQL tuning work requires a high degree of expertise and a massive amount of time, which is the reality of the internal bottleneck faced by organizations in the AI transition period.

Openmade Consulting has directly experienced and researched this on-site problem for over 20 years. The result of that is 'QueryMedic,' an AI LLM-based SQL auto-tuning solution.

[QueryMedic: AI LLM-Based Autonomous Database SQL Optimization Platform]

QueryMedic is not just a simple AI SQL tuner. It is a database SQL performance optimization platform based on an AI Co-pilot that implements an End-to-End auto-tuning pipeline leading from Generate → Analyze → Rewrite → Validate → Predict Impact.

It is designed to directly overcome the core flaw of AI pointed out by the FT: ‘mistaking an increase in activity for value creation.’ The value of QueryMedic is proven not by the number of completed tasks, but by a single metric: performance improvement.

[Conditions for Breaking the Productivity Paradox: Immediately Measurable ROI Without Organizational Redesign]

There are two core conditions of the productivity paradox pointed out by Burn-Murdoch.

First is when technology is simply placed on top of existing work methods. Second is when performance is not measured. QueryMedic fundamentally resolves both of these conditions.

• General AI Apps - App launches increase explosively, but real users do not increase. Only 'activity' increases.

• AI Coding Assistants - Code writing speed improves. However, bugs, review, and verification costs are difficult to measure separately.

• QueryMedic - Presents the SQL tuning effect numerically and automatically verifies whether there are errors in the execution result values of the tuned SQL.

QueryMedic does not change the workflow and role of the existing DBA (performance manager). AI provides the analysis, tuning, and results, and the DBA decides the judgment and application. This is the exact opposite approach to the ‘failure of placing technology on the existing layout without complementary investment’ mentioned by Burn-Murdoch. The DBA's expertise is maintained, and AI takes charge of the repetitive and time-consuming process of searching for DB performance failure factors and improving performance (tuning). The results are immediately confirmed numerically.

[The Top 20% Monopolize Value — Why Companies Operating DBs Must Choose It]

According to PwC's 2026 AI Performance Study, the top 20% of companies take 74% of the economic value generated by AI. The cause of this gap is not the scale of investment. It is whether AI was applied to growth rather than efficiency, and whether it has a measurable foundation.

In all enterprises operating RDBMS-based systems, query performance is not just an IT operation metric. The core flow of business, such as real-time transaction processing in finance, responsiveness of civil complaint systems in public institutions, and ERP/SCM processing speed in manufacturing and distribution, is directly linked to DB query performance.

Only organizations that shorten the SQL tuning cycle, reduce system load, optimize DB operation resources, concentrate DBA capabilities on value-centric strategic tasks, and increase App development productivity with QueryMedic can enter the top 20% of AI value.

Now, the market's evaluation criteria are also changing. It is shifting from ‘who possesses larger technology’ to ‘who proves business value most efficiently.’ The method of blindly expanding infrastructure and manpower, thereby inviting cost pressure, is no longer sustainable. Ultimately, the success or failure of a company in the AI transition period depends on pulling the efficiency of internal resource utilization to the limit to secure substantial cost control and zeroing service failures in information systems. DB performance optimization is the starting point.

[Conclusion: The Value of AI Exists Only When It Is Measured]

FT John Burn-Murdoch's awareness of the problem is accurate. AI increases activity explosively, but if that activity is not converted into value, it will merely become another chapter of the productivity paradox. QueryMedic concretely resolves this paradox in the measurable area of DB performance through an End-to-End pipeline of Generate → Analyze → Rewrite → Validate → Predict Impact. These figures are the evidence of the value that QueryMedic, an AI Co-pilot for DB SQL performance optimization engineering, actually creates in the practical field.

Sales inquiries for product implementation and consulting

02-6310-6167 / qm@openmade.co.kr

Source: IT DAILY ( http://www.itdaily.kr/news/articleView.html?idxno=239842 )

#SQLTuning #AITuning #DatabasePerformance #QueryMedic #OpenmadeConsulting #DBOptimization #AutomationSolution #2025NewProduct #DBA #Developer #ITOperations #CostReduction #PerformanceImprovement #AIAutomation #DBPerformanceManagement #InfrastructureCostReduction