On Thursday 22 October the next edition of the great event Data Restart 2025 took place, focusing on various topics from data and web analytics, BI and AI. This year’s Data Restart had several talks from different areas of data analytics, from the principles, benefits and architecture of server-side tracking, through the use of AI in analysing questionnaire survey data, to modern tools for building your own AI agents. The whole event was, as usual, accompanied by the pleasant moderation of Michael Koch.
Archetix was represented at the conference by Pavel Košan, our great data analyst, who can conjure outputs with big added value from data. A summary of his insights:
Current Analytics Trends That Will Become a Necessity in Time
Speaker: Roman Appeltauer
The series of talks was opened by Roman Appeltauer of Signals, who focused on identifying current technology trends with the greatest potential for long-term survival. He pointed out that most trends turn out to be fleeting over time. It is therefore good to have an idea of which trends will most likely remain present.
Among the current trends with this potential Roman named above all SGTM (server-side measurement), composable architecture (that is, the ability to assemble a solution from various parts and components rather than buying one big colossal system for everything) and GCP (Google Cloud Platform), which is an ever bigger part of the technology stack of many companies and does not look like this trend is weakening.
Seznam Event Measurement: A New Way of Measuring Retargeting and Conversions
Speakers: Jakub Drahokoupil, Nikola Dvorščáková
Next the floor was taken by Jakub Drahokoupil and Nikola Dvorščáková from Seznam, who presented SEM (Seznam Event Measurement), a new way of measuring interactions that will replace the current Sklik conversion measurement. Some of SEM’s features include a unified measurement code, server-side measurement, more first-party data, and advanced retargeting and micro-conversions. Internal testing is still under way and the overall release is expected in the first quarter of next year (2026).
From a Hundred Interviews to an Answer: How to Tame Research Data with AI
Speaker: Gabriela Ostárková
An interesting talk was offered by Gabriela Ostárková from Heureka, who works on UX research. UX research (user research) is an area that deals with uncovering human behaviour, moods and needs and translating this knowledge into marketing, product design and so on. UX research methods include in-depth interviews and questionnaire surveys. Gabriela focused on how AI currently helps her with some aspects of her work. From automatic speech-to-text conversion, through AI-enhanced analysis of interviews, to analysing questionnaires or building a research repository. Today’s AI is very useful for many activities in UX research and can make a lot of tedious work more efficient, letting the UX researcher concentrate on activities that really have an impact.
Using Purchase Data for Strategic Planning and Hyper-Personalisation of Marketing
Speaker: Radomír Dolejší
Radomír Dolejší, a manager at Visa, showed us what opportunities there are in analysing purchase data from credit cards. Our purchasing behaviour can say a lot about us as individuals and as a society. And this knowledge can be very useful for strategic planning, whether for market segmentation or the strategy of marketing campaigns. In general, payment transaction data is a powerful tool for deep insight into consumer behaviour and the economy.
AgentSpace: It Does More So You Can Do Less – Are You Ready?
Speaker: Pavol Lukáč
A very beneficial talk was the one by Pavol Lukáč of TRKKN, who presented AgentSpace. It is a platform for businesses from Google Cloud that lets organisations adopt and scale the use of AI agents. Among other things, AgentSpace helps employees search, summarise and act using company knowledge scattered across various systems (such as Drive, Confluence, Jira, BigQuery and others); it offers unified, secure search across the whole organisation that is based on artificial intelligence and built on Google Cloud’s enterprise infrastructure with an emphasis on data security, privacy and compliance.
Build Your Own AI Agent on Top of Company Data
Speaker: Ivan Kutil
Ivan Kutil, a leading Czech expert on Google Cloud and co-founder of the company AppSatori, told us what AI agents are, what components make up such an AI agent, and presented the basic pattern of the ReAct agent. He then introduced an interesting tool, ADK (Agent Development Kit), Google’s tool for building AI agents (something like Google’s LangChain). He showed how ADK works and how to build and deploy your own simple AI agent in it. We then also learned basic information about MCP (Model Context Protocol), an open standard that allows AI systems (such as large language models) to communicate with and access external tools, data sources and services in a standardised way, and so give them context for performing actions.
Server-Side Stack
Speaker: Jan Tichý
Jan Tichý from Taste presented the concept of server-side measurement. He showed us the whole evolution of web measurement from the beginnings with native scripts in the browser up to modern server-side solutions, and showed us the complete modern architecture for web and mobile analytics including integration with CDP/CRM/ERP systems. We use this infrastructure with most of our key clients, both for supplementing data into GA4, for setting up conversions, and for communication with mailing platforms. It turned out that every company approaches the issue differently, and so the implementation of a server-side stack can differ quite a lot between clients.