Author: IMASTUM

Method
Why AI-agent analytics still needs a human: the foreign-agent case
Where our “12-minute” analytics still needs a human: a missed error, the architecture of verification and responsibility for the findings.
Read the article
AI under scrutiny
Digital avatars and trust: what the Tavus Griffin experiment showed
Examining the limits of what a video-avatar experiment can tell us.
Read the article
Research
EEF 2026 on Telegram: anatomy of an information tsunami
Updated demonstration Brief: 4,592 posts, around 136 million views and seven source types. PDF and interactive report.
Read the article
Method
Verify and compare: how we work with sources
Why several AI answers do not necessarily amount to independent confirmation.
Read the article
Research
RUB 193 billion: how the HoReCa risk estimate works
A full explanation of the scenario model: the revenue-at-risk formula, a restaurant example and answers to questions about quality, coefficients and the calculation’s limits.
Read the article
AI under scrutiny
“What is your wish, Your Excellency?” How AI mastered the oldest of arts
Why AI agrees with its user, how flattery reinforces misconceptions and why the digital courtier is more dangerous than it seems.
Read the article
Reviews and sales
How review monitoring can help increase sales
How to collect user reviews from different sources and use them on an online store’s product pages.
Read the article