For decades, standardized industry codes have formed the backbone of economic analysis. Germany’s Economic Activity Classification (WZ 2008) and international standards such as the NACE codes make it possible to systematically record economic activities and gain aggregated insights into economic sectors. Whether to determine the number of companies in manufacturing or to analyze macroeconomic trends—traditional industry codes hold a firm place in economic statistics.
Yet this apparent strength is increasingly revealing itself as a fundamental weakness: the rigidity of these systems can no longer keep pace with the dynamics of modern markets.
The shortcomings of traditional industry codes become particularly evident in the granularity of classification in emerging economic sectors. While agriculture has numerous highly detailed codes—for example, for the cultivation of specific legumes—the IT sector is reduced to only a handful of broad categories.
The result: app developers are lumped together with providers of highly specialized industrial software under a single code. A differentiated analysis of disruptive subsectors becomes impossible. Anyone conducting market analyses or making strategic decisions today encounters systematic blind spots precisely where innovation takes place.
Foundational technologies, such as artificial intelligence, can permeate nearly all areas of the economy. This horizontal diffusion is, by definition, not captured by vertical industry codes. The consequence: decision-makers in business, research, and policy cannot determine how many AI companies exist, nor can they differentiate their specific technologies and application fields.
Even in traditional industries, industry codes reach their limits when detailed market insights are required. A mechanical engineering company can be identified—but a manufacturer of packaging machines for bamboo materials cannot. The necessary granularity simply does not exist in current classification systems.
Companies continuously adapt their portfolios to changing markets. Traditional industry codes cannot track this development because they rely on static assignments. A company originally focused solely on mechanical engineering that now also offers innovative AI-integrated solutions does not differ—according to its industry code—from a competitor that does not.
ISTARI addresses this challenge with a fundamental paradigm shift: instead of rigid codes, current activity profiles are generated in real time through AI-agent analysis workflows. Autonomous AI agents analyze all publicly available sources, including:
• Company websites with up-to-date product and service information
• Job postings as indicators of technological orientations
• Annual reports containing strategic insights
• News articles and executive interviews
• Registers and databases as complementary sources
Based on this, dynamic company profiles emerge that precisely reflect current activities—independent of historical industry classifications. The contents of these profiles can be tailored to the specific use case. This enables companies and markets to be identified and monitored according to any desired criteria.
Traditional industry codes will not disappear entirely. For highly aggregated macroeconomic analyses and long-term time series, they retain their value.
However, the future belongs to hybrid approaches: while industry codes map the macro level, AI-agent systems enable highly individualized search logics for fine-grained analyses. Companies and organizations can be identified through precise characteristics, whether specific product features, technology use, or network integration.
This flexibility is not a technological gimmick but an economic necessity: in an era of accelerated innovation and fluid market boundaries, decision-makers require analytical tools capable of keeping pace with reality.