In this conversation
The context behind the clip.
Brandon Toombs joins Imran Sajid, Raj Sharmacharya, and Todd Asevedo to discuss what he has actually built with AI. The examples move from Employee Central data work and project administration to browser-driven tasks in a demo system. The conversation also examines iteration, operating cost, and the changing relationship between administrators and consultants.
For: HRIS administrators, payroll practitioners, and SAP SuccessFactors teams exploring practical AI.
What we get into
- Brandon's examples include an Employee Central conversion tool, a digest of relevant knowledge articles, project tasks from meeting notes, and formatted presentations.
- The crew discusses why a repeatable process matters more than generating another dashboard, and why usage cost belongs in the evaluation.
- Browser-use examples are described in a demo environment with dummy data. They are not evidence that unattended changes to a production HR system are appropriate.
Questions this conversation opens up
How can HR and payroll teams start using AI?
Brandon's examples start with specific jobs: converting Employee Central data, collecting relevant knowledge articles, turning meeting notes into project tasks, and preparing presentations. These are examples from HR technology work, not proof that an AI system can run payroll independently.
Try this with your team: Choose one recurring task whose result you already know how to check. Write down what a good result looks like, use an approved sample, and compare the time spent preparing and reviewing the output with the current process.
What should an AI pilot measure beyond time saved?
The discussion connects repeated refinement with usage cost. A first answer and a dependable workflow are different stages of the work. The cost conversation starts at 34:00; the preceding chapter explores improving an imperfect result.
Try this with your team: Keep a small run log: task completed, review minutes, corrections, usage cost, and whether the output was accepted. Include failed attempts. A faster draft is useful only if the total effort still improves after checking and rework.
Can a browser agent make changes in an HR system?
The episode demonstrates browser-driven HR tasks in a demo environment with dummy data. It shows a type of interaction that can be explored; it does not establish that the same workflow is ready for unattended production use.
Try this with your team: Before a pilot can write records, define the permitted action, the person who approves it, the evidence retained, and how a mistake would be corrected. A useful first boundary is to prepare a proposed change for a person to review.
Bring it back to your team
The useful starting point is a task you understand well enough to check. Bring its inputs, expected output, review step, and actual cost into the conversation. The worksheet below turns that into a small, testable pilot.
- Which recurring task has a clear, checkable result?
- What changes when a prototype becomes something a team depends on?
- Who reviews an action before it changes an HR or payroll record?
From the conversation
Brandon Toombs on getting started
I used AI to create a conversion tool specifically designed for Employee Central with position management. Another thing I did: I used AI to build me a customized daily report of all of the KBAs and blogs in the EC area so I can stay on top of the news. Next, I used AI to turn meeting summaries into tasks for project logs, and to create PowerPoint presentations in the customer's format.
