Last CallLast Call Pool

Design Stage

Modern content design team: human expertise AI can’t replace

September 17, 2025 9:40 AM

Long Lecture

🇬🇧 English

Wisła

About

- The evolving role of content design competence: how AI influences our responsibilities, and what skills are in demand

- Future-proofing your team’s skills: shifting to strategic work within design competence from IC content delivery approach

- Building leadership skills: teaching ICs how to influence product direction in cross-functional initiatives (especially important in automation era)

- Keeping your craft alive: how to keep your team motivated, inspired, and creatively open in era of automation

Talk by

Photo of Oleksii Tkachenko, Senior Content Designer at TravelPerk

Oleksii Tkachenko

Senior Content Designer at TravelPerk

Watch the full talk

Watch this WaysConf session, then continue with related talks or explore the current programme.

Explore the 2026 agenda More WaysConf recordings

From the recording

Talk in brief

In this talk from WaysConf 2025, Oleksii Tkachenko, Senior Content Designer at TravelPerk, discusses the evolving role of content design in the age of artificial intelligence. Drawing from his experiences, Tkachenko contrasts organizations that view AI as a cost-cutting replacement for human workers with those that empower designers to own and direct AI tools. He argues that while AI excels at basic copy drafting, localization, and grammar checks, it remains incapable of solving complex user experience (UX) problems, aligning cross-functional teams, or maintaining true consistency with core design principles. Through real-world examples from TravelPerk—such as resolving a confusing hotel virtual payment flow and auditing a generative chatbot named Juno—Tkachenko demonstrates that human judgment, strategic collaboration, and rigorous output evaluation are irreplaceable. Ultimately, he outlines a three-tiered support model and three pillars of modern content design (system building, strategic collaboration, and evaluation) to show how AI allows designers to shift from manual execution to high-impact decision-making.

Key takeaways

01
AI Ownership Empowers Designers
When design teams own and direct AI tools rather than competing against them, they retain strategic influence and can focus on high-level craft.
Watch from 2:20
02
AI Struggles with Complex UX
AI tools tend to generate literal, biased solutions based on user prompts rather than uncovering the underlying user experience problem.
Watch from 5:15
03
The Pitfall of AI Self-Criticism
Asking an AI to critique its own output does not yield an objectively better result. Instead, it merely produces a different or restructured response to satisfy the user's prompt.
Watch from 13:24
04
Three-Tiered Support Model
Content design teams can scale their impact by establishing self-serve, lightweight, and full-support workflows based on project complexity.
Watch from 17:28
05
Redefining Junior Roles
The next generation of junior content designers will not just write copy, but will be native AI operators who know how to integrate these tools into the design process.
Watch from 28:12

Video chapters

0:00Two Approaches to AI Integration
Oleksii contrasts using AI as a cost-cutting replacement for human designers with using it as an empowering tool owned by the design team.
4:12What AI Can and Cannot Do
An overview of AI's strengths in basic copy drafting and localization, alongside its limitations in solving complex UX problems.
5:28Case Study: The Hotel Payments UX Challenge
A real-world example of how an AI-generated solution failed to address the actual traveler experience during hotel check-ins.
9:42Why AI Fails at Self-Evaluation and Alignment
An analysis of AI's susceptibility to user bias, its tendency to hallucinate UX principles, and its inability to align cross-functional teams.
17:20The Three Levels of Content Design Support
A framework for organizing content design resources into self-serve, lightweight, and full-support tiers.
20:18Three Pillars of Modern Content Design
How modern content designers must focus on system building, strategic collaboration, and continuous evaluation.
23:34Case Study: Auditing the Juno Chatbot
A humorous yet cautionary example of a generative chatbot failing to understand basic user inputs, highlighting the need for human validation.
25:28The Future of Design Boundaries
A look at how design roles are expanding, with product designers experimenting with code and junior roles being redefined around AI fluency.

Read edited transcript highlights

These concise notes were edited from automatic captions and checked against the talk structure. They are not a verbatim transcript.

Surviving AI vs. Owning the Tool

In some organizations, AI is introduced as a cost-cutting measure to replace staff, forcing designers to constantly defend their expertise against stakeholders who recheck everything with ChatGPT. At TravelPerk, the approach is different: designers own the AI tools, deciding where they fit in the workflow to automate tedious manual tasks while preserving human strategic influence.

Watch from 0:00

The Hotel Payments UX Pitfall

When TravelPerk faced a complex issue with indirect hotel payments and virtual cards, a product manager asked an AI to write an explanation. The AI generated a long, technically accurate description of the business relationship. However, this missed the actual user need, which was knowing where to find the card and how to use it at check-in. A human content designer reframed the problem to focus on the check-in journey rather than the backend business model.

Watch from 5:28

The Illusion of AI Self-Correction

Many product managers believe they can refine AI outputs by asking the tool to act as a senior critic and revise its own work. This is a misunderstanding of how LLMs operate. Because the AI's primary objective is to please the user, asking it to criticize itself will always result in changes, but these changes are often just different phrasing rather than an objective improvement.

Watch from 13:00

Implementing a Labeled Self-Serve Model

To scale content design, teams can implement a self-serve tier where product managers and designers use internal AI tools for minor copy changes. Crucially, any project completed this way must be labeled as AI-involved. This labeling allows the content design team to audit the output later and protects them from being blamed for poor automated solutions they did not review.

Watch from 17:37

The Three Pillars of the Modern Content Designer

The modern content designer's role is built on three pillars: system building, strategic collaboration, and evaluation. System building involves training internal AI tools and maintaining prompt libraries. Strategic collaboration means aligning teams and leading content-heavy projects. Evaluation requires running continuous audits to ensure AI-generated copy meets the company's quality standards.

Watch from 20:18

The Failure of Generative Chatbot Responses

During an experiment where TravelPerk enabled generative AI responses for their chatbot, Juno, a user tried to book a train from New York to Boston using the abbreviations NY and BO. The chatbot got stuck in a loop asking if the user meant Boston in the US or UK, and then failed to understand the response US, asking if the user wanted travel plans for us. The user rage-quit, proving that automated systems still require strict human guardrails and evaluation.

Watch from 23:34

Explore WaysConf 2026

View the agenda

Get tickets