How Role Context Transforms AI Results in Workday

Remember how we talked about starting with objects? Well, here's what happened when I learned there's something just as important as that.

Last month, I was tasked with helping deploy a new bonus plan. Simple enough concept – roll out bonuses to U.S. employees who've been with us at least 90 days. Finance wanted the first run in April, which meant if I didn't get this configured in time, we'd either delay bonus payouts or scramble with manual workarounds.

Sounds straightforward, right?

The problem that wasn't in the requirements

The eligibility rules seemed simple when the compensation team first explained them. But once I started digging into the actual Workday setup, I realized we had some messy data situations:

Some people had wrong Primary Locations (thanks to a rushed onboarding processes)
We had re-hires where the Hire Date reset but their Continuous Service Date was what we actually cared about
A few people had future-dated position changes that were complicating things

So, I needed a calculated field that would return TRUE for bonus eligibility, but it had to handle all these edge cases cleanly. No pressure.

My first attempt generated a response that was too comprehensive

I opened up Mando AI and typed what felt like a reasonable prompt to get started thinking through it. I figured I could iterate my way to an answer.

"Help me build a calculated field for bonus eligibility based on location and tenure."

The response I got back from Mando AI was comprehensive but not exactly what I was looking for. It gave me a very detailed breakdown:

Calculate Tenure: Use Date Difference function from hire date to today, then Evaluate Expression Band or True/False Condition to define tenure ranges
Define Location Condition: Create True/False Condition where "Location equals a specific value"
Combine Conditions: Create final calculated field that returns true when both conditions are met

It even included an incredibly detailed example for calculating "active days" during a fiscal year, involving multiple Evaluate Expression fields, Date Difference calculations, and Leave of Absence handling.

But here's what this response didn’t do for me:

Too generic on location: "Location equals a specific value" doesn't tell me whether to use Primary Location, Country Code, or something else
Defaulted to Hire Date: It didn’t mention that Continuous Service Date might be more appropriate for tenure
Details that weren’t relevant to my needs: The "active days" example was way more detailed than I needed for a simple 90-day check
No best practices: Lots of technical steps but no advice on edge cases or naming conventions

While comprehensive, Mando AI generated a textbook chapter on the fly when what I really needed was someone to say "use this field, avoid that pitfall." That’s when it hit me – it was providing advice as if I was new to Workday rather than an experienced expert.

The problem: I didn't tell the AI who I was

The AI treated my input as a high-level question when I needed more tactical help building a configuration. It was responding to me like someone who needed to report on something, not someone who needed to actually configure something new in the tenant.

So, I used a different approach. Instead of adding a lot of detail, I just added one thing - my role:

"As an HRIS Analyst, help me build a calculated field for bonus eligibility based on location and tenure."

The difference was night and day

Just by adding "As an HRIS Analyst" to the beginning, the response from Mando AI completely changed.

Why role context changes everything

Here's what I learned: when you tell the AI what your role is, you're not just giving it information - you're changing how it categorizes your entire question.

Without role context, Mando treated my question as a reporting problem - it focused on complex calculations, fiscal year considerations, and detailed data manipulation.

With role context, Mando recategorized the same question as an HCM configuration problem - suddenly it was focused on practical field types, implementation steps, and how to structure the logic in Workday.

Same question, completely different lens.

Generic prompt gets you: Conceptual framework and examples of what’s been tried before
Role-specific prompt gets you: Step-by-step implementation with specific calculated field types

I've started using this pattern for everything now:

"In Workday, as a [YOUR ROLE], [ACTION] on the [OBJECT] that [SPECIFIC LOGIC/GOAL]..."

How it played out

Once I implemented the calculated field using that refined guidance from Mando AI, everything fell into place:

The field worked correctly on the first test run
Edge cases were handled cleanly
The downstream eligibility rule configuration was straightforward
Finance got their bonus accrual data ahead of schedule

Most importantly, I didn't waste a week going back and forth with vague responses that weren't actionable. That refined guidance from Mando AI wasn't just helpful - it was production-ready.

Your turn to try this

Next time you're prompting any AI tool about Workday, start by telling it exactly who you are:

“As a HRIS Analyst...”
“As a Compensation Specialist...”
“As a Security Administrator...”
“As a Integration Developer...”
“As a System Administrator...”

Then be ruthlessly specific about what you're trying to build, configure, or solve.

The difference in response quality is dramatic. You'll go from high-level advice to step-by-step instructions you can build from.

What's next

In our next post, we'll be diving into something that completely changed how I write prompts: Workday-native action verbs.

But first - try the role context trick on your next prompt. Seriously, just add "as a [your role]" to the beginning and watch what happens.

Remember how we talked about starting with objects? Well, here's what happened when I learned there's something just as important as that.

Last month, I was tasked with helping deploy a new bonus plan. Simple enough concept – roll out bonuses to U.S. employees who've been with us at least 90 days. Finance wanted the first run in April, which meant if I didn't get this configured in time, we'd either delay bonus payouts or scramble with manual workarounds.

Sounds straightforward, right?

The problem that wasn't in the requirements

The eligibility rules seemed simple when the compensation team first explained them. But once I started digging into the actual Workday setup, I realized we had some messy data situations:

Some people had wrong Primary Locations (thanks to a rushed onboarding processes)
We had re-hires where the Hire Date reset but their Continuous Service Date was what we actually cared about
A few people had future-dated position changes that were complicating things

So, I needed a calculated field that would return TRUE for bonus eligibility, but it had to handle all these edge cases cleanly. No pressure.

My first attempt generated a response that was too comprehensive

I opened up Mando AI and typed what felt like a reasonable prompt to get started thinking through it. I figured I could iterate my way to an answer.

"Help me build a calculated field for bonus eligibility based on location and tenure."

The response I got back from Mando AI was comprehensive but not exactly what I was looking for. It gave me a very detailed breakdown:

Calculate Tenure: Use Date Difference function from hire date to today, then Evaluate Expression Band or True/False Condition to define tenure ranges
Define Location Condition: Create True/False Condition where "Location equals a specific value"
Combine Conditions: Create final calculated field that returns true when both conditions are met

It even included an incredibly detailed example for calculating "active days" during a fiscal year, involving multiple Evaluate Expression fields, Date Difference calculations, and Leave of Absence handling.

But here's what this response didn’t do for me:

Too generic on location: "Location equals a specific value" doesn't tell me whether to use Primary Location, Country Code, or something else
Defaulted to Hire Date: It didn’t mention that Continuous Service Date might be more appropriate for tenure
Details that weren’t relevant to my needs: The "active days" example was way more detailed than I needed for a simple 90-day check
No best practices: Lots of technical steps but no advice on edge cases or naming conventions

While comprehensive, Mando AI generated a textbook chapter on the fly when what I really needed was someone to say "use this field, avoid that pitfall." That’s when it hit me – it was providing advice as if I was new to Workday rather than an experienced expert.

The problem: I didn't tell the AI who I was

The AI treated my input as a high-level question when I needed more tactical help building a configuration. It was responding to me like someone who needed to report on something, not someone who needed to actually configure something new in the tenant.

So, I used a different approach. Instead of adding a lot of detail, I just added one thing - my role:

"As an HRIS Analyst, help me build a calculated field for bonus eligibility based on location and tenure."

The difference was night and day

Just by adding "As an HRIS Analyst" to the beginning, the response from Mando AI completely changed.

Why role context changes everything

Here's what I learned: when you tell the AI what your role is, you're not just giving it information - you're changing how it categorizes your entire question.

Without role context, Mando treated my question as a reporting problem - it focused on complex calculations, fiscal year considerations, and detailed data manipulation.

With role context, Mando recategorized the same question as an HCM configuration problem - suddenly it was focused on practical field types, implementation steps, and how to structure the logic in Workday.

Same question, completely different lens.

Generic prompt gets you: Conceptual framework and examples of what’s been tried before
Role-specific prompt gets you: Step-by-step implementation with specific calculated field types

I've started using this pattern for everything now:

"In Workday, as a [YOUR ROLE], [ACTION] on the [OBJECT] that [SPECIFIC LOGIC/GOAL]..."

How it played out

Once I implemented the calculated field using that refined guidance from Mando AI, everything fell into place:

The field worked correctly on the first test run
Edge cases were handled cleanly
The downstream eligibility rule configuration was straightforward
Finance got their bonus accrual data ahead of schedule

Most importantly, I didn't waste a week going back and forth with vague responses that weren't actionable. That refined guidance from Mando AI wasn't just helpful - it was production-ready.

Your turn to try this

Next time you're prompting any AI tool about Workday, start by telling it exactly who you are:

“As a HRIS Analyst...”
“As a Compensation Specialist...”
“As a Security Administrator...”
“As a Integration Developer...”
“As a System Administrator...”

Then be ruthlessly specific about what you're trying to build, configure, or solve.

The difference in response quality is dramatic. You'll go from high-level advice to step-by-step instructions you can build from.

What's next

In our next post, we'll be diving into something that completely changed how I write prompts: Workday-native action verbs.

But first - try the role context trick on your next prompt. Seriously, just add "as a [your role]" to the beginning and watch what happens.

Remember how we talked about starting with objects? Well, here's what happened when I learned there's something just as important as that.

Last month, I was tasked with helping deploy a new bonus plan. Simple enough concept – roll out bonuses to U.S. employees who've been with us at least 90 days. Finance wanted the first run in April, which meant if I didn't get this configured in time, we'd either delay bonus payouts or scramble with manual workarounds.

Sounds straightforward, right?

The problem that wasn't in the requirements

The eligibility rules seemed simple when the compensation team first explained them. But once I started digging into the actual Workday setup, I realized we had some messy data situations:

Some people had wrong Primary Locations (thanks to a rushed onboarding processes)
We had re-hires where the Hire Date reset but their Continuous Service Date was what we actually cared about
A few people had future-dated position changes that were complicating things

So, I needed a calculated field that would return TRUE for bonus eligibility, but it had to handle all these edge cases cleanly. No pressure.

My first attempt generated a response that was too comprehensive

I opened up Mando AI and typed what felt like a reasonable prompt to get started thinking through it. I figured I could iterate my way to an answer.

"Help me build a calculated field for bonus eligibility based on location and tenure."

The response I got back from Mando AI was comprehensive but not exactly what I was looking for. It gave me a very detailed breakdown:

Calculate Tenure: Use Date Difference function from hire date to today, then Evaluate Expression Band or True/False Condition to define tenure ranges
Define Location Condition: Create True/False Condition where "Location equals a specific value"
Combine Conditions: Create final calculated field that returns true when both conditions are met

It even included an incredibly detailed example for calculating "active days" during a fiscal year, involving multiple Evaluate Expression fields, Date Difference calculations, and Leave of Absence handling.

But here's what this response didn’t do for me:

Too generic on location: "Location equals a specific value" doesn't tell me whether to use Primary Location, Country Code, or something else
Defaulted to Hire Date: It didn’t mention that Continuous Service Date might be more appropriate for tenure
Details that weren’t relevant to my needs: The "active days" example was way more detailed than I needed for a simple 90-day check
No best practices: Lots of technical steps but no advice on edge cases or naming conventions

While comprehensive, Mando AI generated a textbook chapter on the fly when what I really needed was someone to say "use this field, avoid that pitfall." That’s when it hit me – it was providing advice as if I was new to Workday rather than an experienced expert.

The problem: I didn't tell the AI who I was

The AI treated my input as a high-level question when I needed more tactical help building a configuration. It was responding to me like someone who needed to report on something, not someone who needed to actually configure something new in the tenant.

So, I used a different approach. Instead of adding a lot of detail, I just added one thing - my role:

"As an HRIS Analyst, help me build a calculated field for bonus eligibility based on location and tenure."

The difference was night and day

Just by adding "As an HRIS Analyst" to the beginning, the response from Mando AI completely changed.

Why role context changes everything

Here's what I learned: when you tell the AI what your role is, you're not just giving it information - you're changing how it categorizes your entire question.

Without role context, Mando treated my question as a reporting problem - it focused on complex calculations, fiscal year considerations, and detailed data manipulation.

With role context, Mando recategorized the same question as an HCM configuration problem - suddenly it was focused on practical field types, implementation steps, and how to structure the logic in Workday.

Same question, completely different lens.

Generic prompt gets you: Conceptual framework and examples of what’s been tried before
Role-specific prompt gets you: Step-by-step implementation with specific calculated field types

I've started using this pattern for everything now:

"In Workday, as a [YOUR ROLE], [ACTION] on the [OBJECT] that [SPECIFIC LOGIC/GOAL]..."

How it played out

Once I implemented the calculated field using that refined guidance from Mando AI, everything fell into place:

The field worked correctly on the first test run
Edge cases were handled cleanly
The downstream eligibility rule configuration was straightforward
Finance got their bonus accrual data ahead of schedule

Most importantly, I didn't waste a week going back and forth with vague responses that weren't actionable. That refined guidance from Mando AI wasn't just helpful - it was production-ready.

Your turn to try this

Next time you're prompting any AI tool about Workday, start by telling it exactly who you are:

“As a HRIS Analyst...”
“As a Compensation Specialist...”
“As a Security Administrator...”
“As a Integration Developer...”
“As a System Administrator...”

Then be ruthlessly specific about what you're trying to build, configure, or solve.

The difference in response quality is dramatic. You'll go from high-level advice to step-by-step instructions you can build from.

What's next

In our next post, we'll be diving into something that completely changed how I write prompts: Workday-native action verbs.

But first - try the role context trick on your next prompt. Seriously, just add "as a [your role]" to the beginning and watch what happens.