People and Algorithms at PUEB: marketing automation, data security, and practical AI
Author: Wojciech Patelka
Two days about making AI work inside organizations
On 23 and 24 June 2026 I took part in People and Algorithms: What It Takes to Make AI Work in Organizations? at Poznan University of Economics and Business. The conference was part of the DIGIT project, “People and algorithms in organizations: competencies to work in the digital environment”, which focuses on the skills needed for digital work and Marketing 5.0.
That framing suits the topic. Conversations about AI drift toward tools, demos, and big predictions. Inside organizations, progress depends on three plainer things: the process, the data, and people who understand what the tool does, where it helps, and where it should not be trusted without a check.
Workshop: marketing automation
On the first day I ran the Polish-language workshop Automatyzacja w marketingu for registered students. I gave participants free access to Claude and Claude Design, and we used Claude Design to prototype a brand design. Working on a concrete task kept the session practical. I did not want a “magic prompt” show. I wanted participants to leave with a way of thinking they could use later in a real team.
A brand prototype is a good test case for marketing automation, because it touches every part of a marketing process: a brief, a draft, a review, and an approval. A model can produce the draft in minutes. The brief and the approval still need people, and the workshop made that visible in practice.
We also looked at where automation makes sense and where it only moves chaos from one tool to another. A good automation starts with a repeatable process that has five parts: input, decision, responsibility, output, and quality control. Choose the tool after you can describe those five parts for your own team.
A template for describing a process
Before automating anything, write one short description of the process with these fields:
- Input: what arrives, from whom, and in what form.
- Decision: who or what decides what happens next, and by which rule.
- Responsibility: who answers for the result.
- Output: what leaves the process, and where it goes.
- Quality control: how you check the output, and who does it.
If you cannot fill a field, the process is not ready to automate. That is useful information, and it costs nothing to learn it before you buy a tool.
Main takeaways from the workshop
- Do not automate confusion. If the process is unclear, AI produces unclear results faster.
- Keep a human in the loop where reputation or legal risk appears. Automation can prepare material, but accountability has to stay visible.
- Build small workflows. One stable process beats five impressive demos nobody will maintain.
- Measure time and quality. A useful tool shortens the work cycle or improves the decision. Looking modern is not a result.
Keynote: data security and resilience against disinformation
On the second day I gave the keynote Data Security in Organisations, Building Resilience Against Disinformation. For me it is the natural companion to automation. The more a process depends on data, models, and fast information flow, the more one question matters: how do we know we are working with the right material?
Organizational risk often appears before anyone uses AI badly. It shows up in unverified sources, rushed copying of content, vague access rules, unclear accountability, and work cultures that reward speed over verification.
Resilience against disinformation is therefore an operational competence as well as a media topic. It affects marketing, sales, HR, customer service, and leadership, because each of them makes decisions based on information.
Questions I suggest teams ask
- Where does this information come from, and who can confirm it?
- Who has access to this data, and why?
- What happens if the model output is wrong and nobody notices for a week?
- Who signs off before content leaves the company?
These questions are short enough to fit into a team meeting and concrete enough to produce an action.
Sharing a stage with Prof. Marek Kowalkiewicz
A personal highlight was being in the programme and on the same conference stage as Prof. Marek Kowalkiewicz from Queensland University of Technology. His keynote, Business-to-Algorithm-to-Customer, the Algorithmic Customer Perspective, landed close to the central theme of the event: algorithms are no longer only internal tools. Increasingly, they take part in the relationship between an organization and its customers.
On a personal note, I am almost finished with his book, so listening to the keynote came with a very satisfying extra layer ;D
Meeting people from the academic world
Between sessions I met many academics, including some of my own former lecturers. It was a pleasure to talk with them as a colleague about the topics they once taught me. Conversations like these show how closely practice and research follow each other, and they gave me ideas for the next workshop.
I may return to PUEB for newer projects. If that happens, I will write about it here.
What stays with me after the event
Digital competencies should not be treated as a separate “technology layer”. They are part of normal organizational work. Teams need to understand data, process, automation, security, and accountability, because AI does not replace those topics. It exposes them.
If I had to condense the lesson for companies starting with AI in marketing or operations, it would be this: begin with one concrete process, describe the risks, define the human checkpoint, and only then choose the tools. It is less spectacular than a demo and much closer to business value.
Photos: materials from Poznan University of Economics and Business and the DIGIT project. More context is available on the DIGIT project page and the DIGIT LinkedIn profile.
Want to talk about practical AI in your organization?
I run workshops and implementation work that connect automation, data security, and real team processes. If you want to find where AI can create value without adding chaos, send a note through the contact section.