What if the hardest part of AI at work is not the model, butgetting a clear prompt and a repeatable process before you touch the fancytools?

In this episode of Intelligence Resources, Samantha ViolaWilson and Brittany George sit down with Gabor Kis, Business Systems Analyst in Change Management at DTCC, to walk how a longtime Xerox data architect reskilled into AI, won an internal hackathon on AWS Kiro, and now teachesbusiness builders to treat English as their coding language.

Drawing on his experience spanning Xerox Business Solutionsdata architecture and change and release work at DTCC, Gabor shares the path from Google AI Studio interview prep into a six-week AI hackathon, why his teambet on Kiro, and how prompting plus named beginner-to-agentic stages (Aha, Mirage, AI Vampire, Synergy) beat tool-chasing. He tells the Grafana storybehind an app named Grace that took a week of work to about 45 seconds and helped reclaim roughly 69 weeks.

The conversation covers AWS Kiro specs versus vibe coding,free credits and cost math versus heavy Claude Code bills, Copilot as a prompt workshop, CAB GPT prioritizing hundreds of weekly changes by blast radius, and a personal close: use AI to identify your attachment style and improve.

Connect with Gabor: https://www.linkedin.com/in/gaborkis

Intelligence Resources drops every Tuesday.

CHAPTERS

0:00 Welcome to Intelligence Resources with Samantha ViolaWilson and Brittany George

0:11 Brittany welcomes Gabor through daughter Grace

0:39 From Xerox data architect to DTCC change management

1:37 Google AI Studio interview prep and landing the role

2:05 Betting the hackathon on AWS Kiro

2:55 DTCC scale and why change management cannot fail

3:52 Winning the AI hackathon and starting to teach

5:29 English as the coding language for business builders

5:55 The aha moment and the Grafana project named Grace

7:22 How 45 seconds times 69 spreadsheets saved ~69 weeks

8:05 Hesitant colleagues and the AI adoption gradient

8:48 What is the most important AI skill?

9:00 Samantha and Brittany on prompting, process, andpatience

10:26 Prompting as communication, and why the tool mattersless at first

11:24 Cheat sheet: make the model write a 10-out-of-10prompt

13:02 Probabilistic answers, not perfect deterministic ones

13:56 Coursera tracks and the Vanderbilt prompting course

16:00 Stages: Aha, Mirage, AI Vampire, Synergy

17:41 Nine-agent software factory and Sonny the orchestrator

19:13 OpenClaw experiment, 8.5M tokens, and pulling the plug

21:25 What is AWS Kiro, and why enterprise stacks pickAmazon and Microsoft

23:14 Spec development versus vibe coding

24:37 Free credits, pricing tiers, and small-business costmath

27:09 Copilot aversion, and using Copilot as a promptworkshop

29:58 Change management: build it, they will come, teachteachers

32:20 CAB GPT hackathon deep dive: risk and blast radius

34:52 AI security fears for PHI, SSNs, and financial data

36:17 Post-quantum computing and harvest-now decrypt-later

38:17 What quantum computing means (the maze analogy)

40:04 When quantum and AI meet

42:31 Human in the loop, presence, and empathy

45:05 Always-on work, motherhood, and tech resentment

46:51 New stage: slave to the machine

49:51 Personal AI use: attachment style prompt

51:06 Close, LinkedIn CTA, and Tuesday goodbye

#AI #ChatGPT #AWS #Kiro #PromptEngineering#ArtificialIntelligence #FutureOfWork #ChangeManagement