Back office ops · Production
Grab Integrity Analytics team automates metric reports and fraud investigations with RAG-powered LLMs
The problem
Data Analysts at Grab's Integrity Analytics team faced growing stakeholder demand for data queries, with manual repetitive SQL query writing making analytics slow and inefficient.
Workflow diagram · grounded in source
1
User submits prompt or query
trigger
“Upon a user prompt or query, SpellVault selects the most relevant queries using RAG”
2
RAG selects and executes queries
ai_action
“SpellVault selects the most relevant queries using RAG, executes them and provides a summary of the results to users through Slack”
3
Data-Arks API retrieves data
integration
“Report Summarizer calls a Data-Arks API to generate the data in a tabular format”
4
LLM summarises and generates insights
ai_action
“LLM helps summarise and generate a short paragraph of key insights”
5
Results delivered via Slack
output
“provides a summary of the results to users through Slack”
Reported outcome
Automated report generation saves an estimated 3-4 hours per report, and fraud investigations that were previously time-consuming can now be completed in a matter of minutes via the A* bot.
Reported metrics
Time saved per report3-4 hours per report
Fraud investigation timereduced to a matter of minutes
Reported stack
SpellVaultData-ArksRAGA* botSlackWikiJIRA
Source
https://engineering.grab.com/transforming-the-analytics-landscape-with-RAG-powered-LLM
Read source ↗Frequently asked questions
What did this team achieve with this AI workflow?
Automated report generation saves an estimated 3-4 hours per report, and fraud investigations that were previously time-consuming can now be completed in a matter of minutes via the A* bot.
What tools did this team use?
SpellVault, Data-Arks, RAG, A* bot, Slack, Wiki, JIRA.
What results were reported?
Time saved per report: 3-4 hours per report; Fraud investigation time: reduced to a matter of minutes (source-reported, not independently verified).
How is this back office ops AI workflow structured?
User submits prompt or query → RAG selects and executes queries → Data-Arks API retrieves data → LLM summarises and generates insights → Results delivered via Slack.