Customer support · Production

Empathy uses Guru Knowledge Agents to deliver region-accurate answers to care managers

The problem

When Empathy expanded to Canada, care managers needed region-accurate estate administration information quickly, but the general Guru chatbot pulled from both US and Canadian sources and returned incorrect answers for the user's jurisdiction.

First attempt

The general Guru chatbot mixed US and Canadian content, producing incorrect regional answers. Content cards that listed multiple provinces on a single card further confused the AI, which could not identify which answer belonged to which province.

Workflow diagram · grounded in source
1
Care manager queries via Chrome Extension
trigger
“Empathy embedded the Knowledge Agents directly into the Guru Chrome extension, allowing care managers to quickly access the right information without switching tabs.”
2
Route to regional Knowledge Agent
routing
“They linked each Knowledge Agent to a dedicated US or Canada resource page in Guru, so care managers could quickly navigate to the correct bot based on the user's location.”
3
AI retrieves province- or state-level answer
ai_action
“If a user asked about estate fees in Ontario, the Canada bot would pull the Ontario-specific answer, not a general Canadian one.”
4
Care manager receives accurate answer
output
“Care managers can now find answers quickly and accurately without sifting through irrelevant content.”
5
Content restructured to fix AI confusion
feedback_loop
“After realizing that AI sometimes pulled the wrong answer due to multiple provinces listed on a single card, she worked with Guru's team to restructure the content. She used headers instead of tables and adjusted the layout to make it ea…”
Reported outcome

Regional separation of knowledge bases improved searchability, significantly reduced irrelevant search results, and increased care team trust and confidence in Guru's answers.

Reported metrics
Irrelevant information in searchessignificantly reduced
Questions with no direct answersrarely now get a question where no direct answers yield
Reported stack
GuruKnowledge AgentsGuru Chrome extension
Source
https://www.getguru.com/customers/empathy-accurate-ai-answers
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Regional separation of knowledge bases improved searchability, significantly reduced irrelevant search results, and increased care team trust and confidence in Guru's answers.

What tools did this team use?

Guru, Knowledge Agents, Guru Chrome extension.

What results were reported?

Irrelevant information in searches: significantly reduced; Questions with no direct answers: rarely now get a question where no direct answers yield (source-reported, not independently verified).

What failed first in this deployment?

The general Guru chatbot mixed US and Canadian content, producing incorrect regional answers.

How is this customer support AI workflow structured?

Care manager queries via Chrome Extension → Route to regional Knowledge Agent → AI retrieves province- or state-level answer → Care manager receives accurate answer → Content restructured to fix AI confusion.

Related customer support cases