They debate organisations rushing AI rollouts without learning change-management lessons. They explore when to automate status reporting with LLMs and when that misses the point. They introduce agentic validation for documenting predictions and checking outcomes over time. They unpack monitoring-driven rollouts, rollback safeguards, and iterative AI loops that experiment and refine results.
20:35
forum Ask episode
web_stories AI Snips
view_agenda Chapters
auto_awesome Transcript
info_circle Episode notes
insights INSIGHT
AI Rollouts Are Repeating Old Change Mistakes
Organisations are repeating old change-management mistakes by treating AI rollouts as purely technical deployments rather than learning from past Agile adoption lessons.
Jeffrey Fredrick heard teams focus on automating artifacts (like weekly reports) instead of re-examining the underlying purpose of those artifacts, which risks preserving pointless bureaucracy.
volunteer_activism ADVICE
Reassess Before Automating Status Reports
Do re-evaluate the purpose of existing artifacts before automating them with LLMs rather than just recreating historical reports.
Jeffrey Fredrick warns that automating a status report may remove the human engagement that the report originally enforced.
insights INSIGHT
Use AI For Routine Counting Tasks
Some projects are like brick walls where progress equals simple counts, and computers excel at consistent, repeatable measurement.
Douglas Squirrel argues that for well-understood, countable work, LLMs generating and checking reports can be perfectly appropriate.
Get the Snipd Podcast app to discover more snips from this episode
What’s got Jeffrey - the Founder of CITCON:AI - ruffled about AI?! In this episode, Jeffrey talks about his frustrations with how organisations approach AI rollouts without learning any lessons from the change-management initiatives that have gone before. Join us to get our take on when to hand over reporting to LLMs, when not to, and how to use ‘agentic validation.’
LINKS:
- Alistair Cockburn article: https://ameyakarve.wordpress.com/2012/07/22/characterizing-people-as-non-linear-1st-order-components-in-software-development-cockburn-us/
- IMVU post: http://timothyfitz.com/2009/02/10/continuous-deployment-at-imvu-doing-the-impossible-fifty-times-a-day/
- CiTCON 2026: https://citconf.com/helsinki2026/
--------------------------------------------------
You'll find free videos and practice material, plus our book Agile Conversations, at agileconversations.com
And we'd love to hear any thoughts, ideas, or feedback you have about the show: email us at info@agileconversations.com
--------------------------------------------------
About Your Hosts
Douglas Squirrel and Jeffrey Fredrick joined forces at TIM Group in 2013, where they studied and practised the art of management through difficult conversations. Over a decade later, they remain united in their passion for growing profitable organisations through better communication.
Squirrel is an advisor, author, keynote speaker, coach, and consultant, and he's helped over 300 companies of all sizes make huge, profitable improvements in their culture, skills, and processes. You can find out more about his work here: douglassquirrel.com/index.html
Jeffrey is Vice President of Engineering at ION Analytics, Organiser at CITCON, the Continuous Integration and Testing Conference, and is an accomplished author and speaker. You can connect with him here: www.linkedin.com/in/jfredrick/
Troubleshooting Agile