ZARZA We are Zarza, the prestigious firm behind major projects in information technology.
![Talk IT to Me [Bonus Episode] AI Agents: between hype and real value](https://i1.sndcdn.com/avatars-Gjwv2rAEZGJZnQJW-Mu18AA-original.jpg){kind=avatar}
Listen "Talk IT to Me [Bonus Episode] AI Agents: between hype and real value"
Episode Synopsis
𝐇𝐨𝐬𝐭𝐬/𝐄𝐱𝐩𝐞𝐫𝐭𝐬:
Mateusz Kaczorek, Senior Data Scientist
Karol Kawalerczyk, Senior Data Scientist
𝐀𝐛𝐨𝐮𝐭 𝐭𝐡𝐞 𝐞𝐩𝐢𝐬𝐨𝐝𝐞:
This episode continues the discussion from the Project: AI 2.0 workshops about AI Agents, featuring SoftServe Senior Data Scientists Mateusz Kaczorek and Karol Kawalerczyk. Together, they examine the AI market, clarify what makes an agentic framework, debate multi-agent system design, and compare fine-tuning vs. prompt engineering.
A forward-looking conversation for experts, leaders, and practitioners who want to understand how AI agents can move from hype to measurable impact.
𝐓𝐢𝐦𝐞𝐬𝐭𝐚𝐦𝐩𝐬:
00:00 Intro – AI agents & Project AI workshop
00:35 Guest intro – Karol joins the discussion
01:10 State of AI market – predictions & failures
06:16 Tools & best practices – missing standards
08:00 AI surprises – coffee grinder & Japanese house
11:43 What makes a framework “agentic”?
21:14 Multi-agent systems – when & why to use them
27:07 Framework evolution – agents creating agents
29:19 Rigid vs flexible workflows – design tradeoffs
37:16 Fine-tuning vs prompt engineering
45:42 Evaluating multi-agent systems – path vs regression
53:52 Metrics & business alignment – accuracy first
59:14 Selling AI agents – avoiding “Skynet” promises
1:04:11 Pilot strategy – flashy demos vs back-office wins
1:08:12 Quick-fire game – fine-tune, memory, buzzwords
1:12:46 Risks & guardrails – database & cloud safety
1:14:16 Final remarks – resources & recommendations
Mateusz Kaczorek, Senior Data Scientist
Karol Kawalerczyk, Senior Data Scientist
𝐀𝐛𝐨𝐮𝐭 𝐭𝐡𝐞 𝐞𝐩𝐢𝐬𝐨𝐝𝐞:
This episode continues the discussion from the Project: AI 2.0 workshops about AI Agents, featuring SoftServe Senior Data Scientists Mateusz Kaczorek and Karol Kawalerczyk. Together, they examine the AI market, clarify what makes an agentic framework, debate multi-agent system design, and compare fine-tuning vs. prompt engineering.
A forward-looking conversation for experts, leaders, and practitioners who want to understand how AI agents can move from hype to measurable impact.
𝐓𝐢𝐦𝐞𝐬𝐭𝐚𝐦𝐩𝐬:
00:00 Intro – AI agents & Project AI workshop
00:35 Guest intro – Karol joins the discussion
01:10 State of AI market – predictions & failures
06:16 Tools & best practices – missing standards
08:00 AI surprises – coffee grinder & Japanese house
11:43 What makes a framework “agentic”?
21:14 Multi-agent systems – when & why to use them
27:07 Framework evolution – agents creating agents
29:19 Rigid vs flexible workflows – design tradeoffs
37:16 Fine-tuning vs prompt engineering
45:42 Evaluating multi-agent systems – path vs regression
53:52 Metrics & business alignment – accuracy first
59:14 Selling AI agents – avoiding “Skynet” promises
1:04:11 Pilot strategy – flashy demos vs back-office wins
1:08:12 Quick-fire game – fine-tune, memory, buzzwords
1:12:46 Risks & guardrails – database & cloud safety
1:14:16 Final remarks – resources & recommendations
More episodes of the podcast SoftServe Podcasts
TCP - S3Ep2 - El lado humano del dato
10/06/2025
ПРОДкаст - Е9 - Равновесна точка
05/06/2025
In God we trust