A narrated description can only support an estimated duration because the model must infer how long each step takes. Recording the complete workflow provides an observed duration and makes the baseline more credible.

Treat timing from narrated summaries as an estimate
Record the entire task for an observed baseline
Use video duration as evidence of total execution time
Preserve step-level activity so the AI can map time to work

“Yeah, it's estimated because I go through in my voiceover of it and say what I'm doing.”
08:00

“If you just record an hour-long video of yourself doing something, then it's gonna know it takes you an hour.”
08:30

#time estimation#workflow recording#measurement#ai analysis

Hot Take· 2

Hot Take09:30

Do Not Read Every AI Report—Use It to Build the Next System

A detailed workflow report may be too long to consume manually, especially for a small task. Its greater value can be as structured context for subsequent prompts, agents, workflows, and skills rather than as a document every employee must read.

Treat the report as structured workflow context
Feed it into an LLM to produce targeted prompts
Use its findings to design agents and skills
Avoid forcing people to read unnecessary detail

“I don't think I would ever actually read all these reports, but I would use them as inputs to then build the agents and build…”
Kieran · 10:30

“You're creating like a more detailed schema of what the work is, really.”
Kieran · 10:00

#agents#skills#workflow reports#automation

Hot Take16:00

Knowing the AI Workflow Is Easier Than Changing Human Habits

AI can describe a better process, but that does not ensure people will adopt it. Existing habits pull users back toward familiar tools and routines, while novel capabilities such as coding or analysis may inspire faster adoption because they enable work that was previously impossible.

Separate workflow knowledge from behavioral adoption
Expect people to default to familiar routines
Use newly unlocked capabilities to create motivation
Introduce small changes that can become habits

“I think the big part that AI is going to struggle to help with is behavioral change.”
Kieran · 16:00

“Don't try to get overwhelmed and do all the things, just try to use it in a certain number of small ways and make them…”
Kieran · 17:00

#behavior change#ai adoption#habits#change management

Explainer· 3

Explainer02:30

The Simplest Way to Use an AI Skill Without Installing It

Users do not need a sophisticated agent environment to test a skill. They can upload the skill and its reference files directly into an ordinary LLM conversation, add the source material, and ask the model to run the skill.

Upload the skill and supporting markdown files
Add the transcript or work artifact to analyze
Tell the model explicitly to use the supplied skill
Test the workflow before investing in a permanent installation

“you just literally get like a zip file folder of these markdown files, and you can just upload them to Claude to Chat GPT to…”
02:30

“I literally said run the skill against this transcript.”
03:00

#ai skills#claude#chatgpt#llm workflows

Explainer05:30

What an AI Task Transformation Report Actually Shows

The report reconstructs the current task, tools, pain points, and estimated duration before proposing an AI-assisted alternative. It also estimates time savings, distinguishes efficiency gains from quality improvements, recommends tools, and suggests implementation steps.

Document the current task and tools
Identify pain points in the existing workflow
Compare current and AI-assisted execution times
Estimate both efficiency and quality gains
Recommend tools and immediate next steps

“It understands the tools that are currently used to do this task and the current pain points of doing the task.”
05:30

“Then it gives you a before and after of what are you doing currently and how long it takes versus using AI, how long they…”
06:00

#workflow optimization#time savings#ai tools#process analysis

Explainer12:00

Why Effective Skill Files Need High Instruction Density

The episode describes skill authoring as a constrained design problem because a principal skill file may have a limited line budget. Researching successful public skills and consolidating their strongest patterns can produce a concise template that remains useful across platforms.

Treat the skill file as a constrained resource
Study popular and proven skill structures
Prioritize high-value instructions over verbosity
Move supporting detail into reference files when appropriate
Design for compatibility across multiple AI platforms

“you are limited to 500 lines in one of these files.”
12:00

“the challenge of what you're trying to do is pack the most density possible into 500 lines.”
12:00

#skill files#prompt engineering#agent skills#instruction design

Q&A· 1

Q&A15:00

The Same Workflow Analysis Can Serve Very Different AI Goals

The next action after workflow analysis depends on the desired organizational outcome. The findings might support adoption of a particular assistant, creation of shared skills, automation of a workflow, or development of AI-aware onboarding materials.

Define the transformation goal before acting on the report
Map manual work to a chosen AI platform when adoption is the priority
Create shared skills when repeatability matters
Turn workflow knowledge into onboarding documentation when standardization matters

“It depends what you're trying to do in terms of your AI transformation.”
Kieran · 15:00

“you could create onboarding docs of how we work, and that would actually really help.”
Kieran · 15:30

#ai adoption#employee onboarding#shared skills#change strategy

Tool· 2

Tool01:30

Record Your Work to Reveal Where AI Can Help

A screen recording with spoken narration can give an AI agent enough context to analyze an existing workflow. The agent can then identify repetitive steps, bottlenecks, and opportunities for AI assistance or automation.

Record the complete workflow while performing it
Narrate decisions and actions as they happen
Give the recording transcript to an AI agent
Use the analysis to identify reusable workflows and skills

“you can record yourself just doing work and narrating your work, and you can give that to an agent”
Kieran · 01:30

“that agent will build skills that help you replicate the kind of workflows you're doing on the day-to-day basis”
Kieran · 01:30

#ai transformation#workflow analysis#screen recording#automation

Tool13:00

Use a Skill Architect to Design Better Agent Skills

The showcased meta-skill can create, improve, or organize agent skills while deciding between single-file, multi-file, and skill-family structures. It is intended to encode architectural decisions once so users do not have to reconsider them for every new skill.

Create new skills from a shared template
Refactor existing skills against consistent standards
Choose between single-file and multi-file architecture
Group related capabilities into skill families
Keep skills discoverable, concise, and cross-platform

“It builds remarkable agent skills that are discoverable, concise, effective across all major platforms.”
13:00

“Decide between a single file versus multi-file skill architecture.”
13:00

#skill architect#meta skill#agent design#skill families

Takeaway· 3

Takeaway04:00

Match the AI Transformation Report to the Size of the Work

The demonstrated skill classifies work as a task, individual, team, or department transformation. It changes the depth of its output accordingly, avoiding a department-scale report for a small task while preserving detail for complex organizational work.

Classify the scope before generating recommendations
Use shorter reports for individual tasks
Reserve extensive analysis for team or department transformations
Adjust recommendations to the complexity of the work

“Is this a task transformation, a deeper like individual level transformation, a team level transformation, or a department level transformation?”
04:00

“I have it change the output based on the type of work it is.”
04:00

#scoping#ai transformation#workflow design#reports

Takeaway11:00

Teams Cannot Transform Work They Have Never Clearly Mapped

Complex workflows often span several tools while remaining poorly understood by the team performing them. A detailed analysis creates a shared account of how work currently happens, which the team can validate before agreeing on an improved AI-assisted process.

Map the current workflow across every tool
Make hidden steps and assumptions explicit
Ask the team to validate the map
Agree on the baseline before designing the replacement

“most teams are not clear on like how the work they do actually happens.”
11:00

“And do we agree that that's how it actually happens? And if so, then can we agree on a new and better way to do…”
11:30

#process mapping#team alignment#change management#ai transformation

Takeaway17:00

AI Rewards People with High Agency and Low Tolerance for Bad Work

The hosts argue that effective AI adopters are curious, willing to experiment, and motivated to build what they previously could not. They also have little patience for inefficient processes because AI has reduced the time and difficulty required to improve them.

Act without waiting for perfect instructions
Experiment with newly accessible capabilities
Notice inefficient or low-quality processes
Use AI to improve weak work rapidly
Look for high-agency behavior when evaluating collaborators

“They're super high agency. They're curious. They're out there building.”
17:30

“I suspect the best people you're working with fall into that like high agency, low tolerance bucket.”
18:00

#agency#mindset#ai adoption#performance

Frameworks in this episode

High Agency, Low Tolerance Model

Record-Narrate-Replicate Workflow Capture

Scoped AI Transformation Plan

Skill Architect Method

Small-Way AI Habit Ladder

Transformation Objective Routing

Insights & moments

⚡Myth Buster· 1

AI Workflow Time Estimates Are Better with Full Recordings

◆Hot Take· 2