Overview
Syllabus
My first task was too simplistic
Aaron Levie Tweet, using cloud software
A more complex task and prompt
Transcript Fetching, NotebookLM Usage, Google Drive Usage, Tweet Thread
Demonstration of Operator Session
Operator scrapes Seeking Alpha
Operator creates an organizational folder structure
Skipping permission checks
Operator using Google Drive
Operator searches for HTML to PDF converter
Security risks of Operator using search engine to find tools
Operator finds NotebookLM and uses it to extract insights, creates a podcast
Even though there's no NotebookLM API, we can automate it!
Operator creates a Google Doc
Operator creates a share link I didn't ask it to!
Operator creates a twitter thread with Alphabet earnings insights
Effects on advertising business? Can ads hijack Operator attention?
I've seen Operator get distracted and do something random
Can I just walk away and trust the agent did everything I asked?
Reflections, notes, thoughts on Operator, predictions
No code needed!
No API needed! Just let Operator use the app!
Teaching AI to use software via video?
Dedicated accounts for AI agents?
Iterative agent management, checking in on your agent's progress
Future integration with Scheduled Tasks feature?
App User Experience for agent users?
Security risks, whitelisting websites, organization policies
Github / version control for Operator tasks / recipes?
The diminishing value of writing code by hand
Operator Mobile App?
You may not use Operator, but Operator will use you!
More business use cases, less travel agents please!
Think in terms of chaining SaaS apps
Wrapping Up - I feel like I have a better understanding of agents now
Taught by
Part Time Larry