AI Assistant Basics
Community Assistant
General
Community Assistant & Other Workday Agents
Community Assistant EA Feedback
Community Assistant Future Roadmap
Community Assistant Other
100

This three‑letter approach helps an Assistant pull in external knowledge before generating a response.


What is RAG

Retrieval‑Augmented Generation (RAG) is important for an AI Assistant, because it gives the Assistant access to current or domain‑specific knowledge rather than relying only on what the model learned during training. 

100

These are the goal(s) that a Community Assistant supports Community Members in achieving:


A) Findability - Discoverabillity  - Automating Tasks

B) Replacing Global Search

C) Andrew told us we had to build an Assistant 

D) Automating Tasks


A

What are: Findability - Discoverabillity  - Automating Tasks

100

Yes‑or‑No question:  

The Deployment Agent and the Community Assistant are actually the same thing.


What is NO

The Deployment Agent resides directly within the user's Workday tenant to help them manage the technical side of go-live, such as troubleshooting configuration errors and building out new functional areas. It is automatically available in Customer Central.  To use it in implementation environments (Implementation, Preview, and Sandbox tenants), the user must activate the Deployment Agent Users Domain Security Policy and enable it on the Agent Management Hub through the Agent System of Record (ASOR).

The Workday Community Assistant serves as the user's 24/7 advisor, specifically designed for navigating the Workday ecosystem and helping them maximize their Workday investment. It lives outside of the user's active tenant, and focuses on helping the user get what they need from the Workday Community.

100

These are the early‑access users of the Community Assistant who will help us gather feedback?


A) Workmates only

B) Workmates + Select Customers & Partners

C) Andrew + Sheri + Rob + Aneel

D) Community Workmates Only

B

Who are: All Workmates + Select Customers and a few Partners

100

This is the timeframe for when Community Assistant will be in General Availability for all Community Members?


A) It's already is in GA (no one better pick this option)
B) Tomorrow
C) Early Q3
D) Q4

C

What is Early Q3.

We’re evolving our Early Access program with additional pre‑GA Production code drops and broadening the group of participating customers and partners.


100

This is the time Shilpa & Linda logged off on Assistant Deployment Night - Wednesday May 20th. 


A) 11:50 PM

B) 12:50 AM

C) 2:50 AM

D) The never actually slept

C - What is 2:50am

200

This term refers to the system that learns patterns from data to generate predictions or responses.

What is an AI model 

An AI model is a system that has learned patterns from data so it can generate predictions, decisions, or responses.


Bonus question: What Model does the Community Assistant use?

200

This is the number of teams that actively contributed to building the Community Assistant.

A) 3 

B) 6

C) 11

D) 12

C - What is 11 

Agentforge, Agent Gateway, Venice, Deployment Agent, CIS/ML RAG, Security, BT, Legal, Support, Boldium, COMMUNITY

200

The Community Assistant RAG is updated how frequently?


A) Monthly

B) Weekly

C) Daily

D) Real-Time

B

What is Weekly

Currently, the DA Rag is updated once a week on Thursday Mornings.

200

Over its first nine working days in production, this metric reflects the total number of actions taken and queries submitted within the Assistant.

A) 4,340

B) 7,903

C) 11,870

D) 18,230

C

What is 11,870

200

 This becomes the top priorites for the Community Assistant team after reaching General Availability


A) Assistant Task Completion: Assistant completes Community/CX related tasks and proactively problem solves

B)  Orchestration Vision: Moving toward the CX vision of the Community Assistant acting as a  orchestrator.

C) Nothing

D) A + B 

D

What is Assistant Task Completion + Orchestration Vision

200

What is the number of Security, Legal, Privacy, Middleware Certifications, Platform Security, AI Review, Jira Tickets that required approvals to get Community Assistant to EA.

A) 5

B) 8

C) 15

D) 789

C

What is 15

But felt like 789...........

just ask Leah, Lakshmi, Nick, Sravya, Imaad, Denise, Adam, or any of the other devs who kept getting pinged nonstop to answer the  questions coming....

300

This approach allows us to judge whether the Community Assistant answer is right or wrong.

What is a Golden set of Questions & Answers

Golden set is a human SME verified set of questions and answers used to test accuracy of the Assistant. It acts as the standard reference for evaluating whether responses are correct.

300

Community Assistant runs on what Workday powered Agent platform? 


A) Google AgentSpace

B) AgentForce

C) AgentForge

D) AWS Bedrock Agent

C - What is AgentForge

Community Assistant runs on LangGraph Enterprise Server, internal (Workday) code name Agent Forge 


300

The number Assistants in production for Community today?

A) 0 (nobody better select this!)

B) 1

C) 2

D) 3



C

What is 2

Today there are technically two Assistants in Production Community Assistant (EA) and Support Assistant for our NSCs on the Case create page in Salesforce.


Once we get a technology in place (A2A or MCP) the CX vision is for the Community Assistant to function as the central customer‑facing front door, orchestrating requests and sending to multiple backend CX specialized agents. 

Community Assistant can 1) query these agents as needed, 2)aggregates their responses, and 3)return a combined answer to the customer via the Community Assistant interface.

300

The percentage of EA users who answered Yes to the following question - Where you able to complete your task using the Community Assistant

A) 53%

B) 62%

C) 74%

D) 100%

B


What is 62%


Note: For Customer/Partner Subset only that number goes up to 68%

300

Yes or No - Will Global Search go away with the introduction of the Community Assistant?



What is NO

Search at Workday and in the industry is evolving. The traditional role of traditional search applications (Semantic relevancy engines) have been evolving with LLM powered responses leading the way.  Highly responive, low index time and low cost search appliances are great fits for consolidating near real time data Ïor multi-platform data discoverability and generalized Community insights.

300

The number of new processes & procedures required by Devops to undertake to get Assistant up and running?

A) 2

B) 3

C) 5

D) 7

 C: What is 5


Create alpaca certs, Install Dova,  Agent forge Deployment, Middleware ndoejs Deployment, Create Venice Production SNI

400

True/False: 

LangGraph is the Company behind the agent builder frame work, LangChain, Workday uses to build its Agents and Assistants 


Answer: False 

The Company LangGraph does maintain the LangChain code base, Community(and Workday) leverage the LangGraph framework to build our Assistant. While LangGraph is built on-top of LangChain, LangGraph offers a highly customizable graph/node model for complex traversal pattern for highly customizable workflows.

 

400

These design principles guided the creation of the Community Assistant.


Answer: What are Sana design principles.

Community Assistant largely aligns with SANA design principles, while incorporating adjustments to better support Community UI/UX standards..

400

In order to achieve higher accuracy with Community's RAG tooling Community Assistant implements:  


A) HyDe RAG Generation

B) Agentic Rag Models

C) Semantic ReRanking

D) Multi-shot query generation

E) A&B

F) All of the Above 


F

What is All of the above

400

This metric represents the  percentage of users who return to the Assistant within seven days after their initial engagement.


Answer:

A) 53%

B) 79%

C) 88%

D) 93%

D

What is 93%


Note: Looking at the subset of EA - Customer & Partners the Assistant return rate is 99%!!!!


400

True/False: The Community Assistant is an all-in-one Agent for the Community space.

Answer: What is False

Community's overarchieving goal for our assistant is orchestration; utilizing specialed agents and tools (including additional Community built Agents) that span our Community Ecosystem. To quote Andrew, we are the air traffic controller directing user inputs...

400

This is the combined number of *&!**&! words Nick & Lakshmi dropped during development & testing.


A) Zero: Because they’re professionals… who definitely didn’t write this answer.

B) A number so high the database switched to scientific notation  and......... still overflowed.

C) NONE - because we had the BEST - Dev, QA, Dev-Ops, Comms, PMO, Content, Analytics, Product, Data Warehouse, Release, Leadership team working on this......


Best Answer: C

Let's be for real there were alot of *&!* words dropped by Lakshmi & Nick but we still had THE BEST TEAM EVER!!!!!!!!!!!!

THANK YOU