AI and Human Validation

How does Intratec collect its data?

Data collection draws on official government trade statistics, data published by technology licensors, governmental agencies, and statistical bureaus worldwide, as well as internally developed mathematical models.

All data collection follows a structured monthly cycle. During the last week of each month, data reported by governmental agencies and statistical surveys is gathered. At the beginning of the following month, the collected data is processed, treated, and assessed by our team before being published to subscribers.

The process relies on advanced technologies built into a proprietary computer-based pipeline, developed and maintained by a multidisciplinary team of market experts, computer scientists, and data scientists. This pipeline extracts, stores, processes, and analyzes publicly available data from a wide range of open sources as soon as it is released, transforming raw source data into reliable price and cost information through mathematical models.

Further details about individual source providers are considered trade secrets and are not disclosed per article.

How is the data validated before it reaches subscribers?

All data undergoes rigorous multi-source verification, followed by expert human review, before it is published to subscribers.

Validation is built into the data processing pipeline in two complementary layers. First, automated processing cross-references data from multiple official sources, applying mathematical models that flag inconsistencies and fill gaps when primary source data is not yet available. Second, market experts review model outputs before publication — this human review layer is a stated differentiator that ensures the final data reflects sound, independent judgment rather than automated output alone.

The methodology is grounded in public sources — including international trade statistics, patents, encyclopedias, technical papers, and data from governmental and statistical agencies — and verified against multiple inputs before any assessment is considered final. This multi-source approach is designed to prevent any single source from introducing systematic errors into the published data.

How does automation affect data quality?

Automation improves data quality by reducing manual handling errors and removing the possibility of bias introduced by market participants.

The proprietary computer-based system at the core of data processing extracts, stores, processes, and analyzes official data as soon as it becomes available, enabling consistent and repeatable handling of large volumes of information. Because the system applies the same validated rules every month, it eliminates the variability and risk of error that manual processing would introduce at scale.

Equally important is what automation prevents: by processing data through rule-based, impartial pipelines rather than through analyst judgment calls on individual assessments, the methodology avoids the bias that could arise if assessors had a commercial interest in influencing prices. All assessments are built on official data and processed with auditable methods — the outputs can be independently verified.

Automation and human expertise work together rather than in opposition: automated pipelines handle data extraction and modeling while expert reviewers validate the outputs before publication.

Is price data reviewed by a person before it is published?

Yes. Human review is part of the publication process for all price data.

Our methodology explicitly combines AI tools with expert human review. After the automated pipeline extracts and processes source data into price assessments, the results are reviewed by a team that includes market experts. This review step is the final quality gate before data is made available to subscribers each month.

The combination is deliberate: automation handles the scale and consistency challenges of processing large volumes of official data, while human expertise provides the judgment needed to assess whether outputs meet quality standards before publication.

Are Intratec analysts independent from market participants?

All price assessments are built on official, publicly available data and processed through impartial, auditable methods — independence from producers, traders, and other market participants is a core design principle.

The methodology is deliberately structured to prevent market participant influence. By sourcing data exclusively from government trade statistics, official agencies, and statistical bureaus — rather than from surveys, broker reports, or voluntary submissions by market players — the assessment process has no dependency on information provided by entities with a commercial interest in price outcomes.

The proprietary computer-based pipeline further reinforces this independence: assessments are generated through rule-based mathematical models, not through subjective analyst judgments that could be swayed by market relationships. The published methodology documents are available publicly, making the assessment process verifiable by subscribers and third parties.

How does Intratec detect errors or anomalies in data?

Errors and anomalies are addressed through continuous model reviews, methodology improvements, and a formal assessment complaint process open to subscribers.

On the production side, data quality is maintained through continuous model reviews and periodic methodology updates. When data inconsistencies, distortions, or quality issues are identified internally — whether through updated source data, changes in market conditions, or shifts in methodology standards — assessments may be revised or retired. All such changes are reported monthly in the Intratec Release Notes at medium.com/intratec-release-notes.

On the subscriber side, a formal complaint process exists for cases where published data appears inconsistent with the published methodology. Subscribers who identify a potential error should first contact the Support Team for a detailed explanation of the assessment process. If the concern remains unresolved after that conversation, a formal complaint can be submitted via the Contact Form at intratec.us/help/contact-us, using "Assessment Complaint" as the subject prefix.