Grab is using AI agents to automate analytics workflows, reducing the share of routine work handled by analysts and shortening the time required to answer business questions. The company reports that the share of mechanical tickets handled by analysts fell from 44% in February to 30% in June, covering tasks such as data preparation, alerting, and reporting.
The approach is based on a five-level autonomy model that defines how much of an analytics workflow an agent can own while retaining human oversight. At Level 3, humans frame the question and review the result while agents discover data, write and execute queries, validate results, and draft analysis. At Level 4, agents can plan and orchestrate workflows while humans review defined gates. Level 5 represents end-to-end autonomy, with humans setting objectives, quality thresholds, and escalation rules. Grab retains human accountability for metric definitions, causal interpretation, business assumptions, and final decisions.
Maanas Prabhakar, who leads analytics at Grab, highlighted the changing role of analysts in a LinkedIn post:
The harder question is what an analyst does when an agent handles the data preparation, the analysis, and everything else.
Grab’s Spartan system applies this model to natural language analytics requests, including questions submitted through Slack. The system uses more than 50 skills and 120 analysis frameworks to route requests to specialized workflows. A root cause question can trigger analysis across certified metrics and relevant dimensions, while an experiment question can retrieve an existing scorecard rather than querying the data lake.
Index architecture across our knowledge base (Source: Grab Blog Post)
The company has also invested in the data context required for agents to produce reliable results. Grab maintains more than 5,000 certified tables and metrics, 4,000 context documents, and 2,000 golden records. Its ContextIQ system treats this information as a lifecycle, updating context as instrumentation changes and incorporating fixes identified from production agent failures.
Grab also uses AI agents for analytics operations. Scarlet, for example, handles pipeline failures by performing root cause analysis and can fix failures or escalate them when predefined gates or documented runbooks do not cover the problem.
Scarlet in action on Slack (Source: Grab Blog Post)
For recurring analytics, Grab uses agents to automate metric and OKR commentary, assessing significant movements, breaking them down across countries and segments, and correlating them with operational changes and experiments. Its BriX portal supports analytics workflow development, with usage growing more than tenfold since September. The team reported 31 production deployments, 283 merge requests, and 60 features during the first half of the year.
Between March and May, self-service analytics answered without human involvement increased from 53% to 67% for metric requests, 63% to 90% for data pulls, and 50% to 81% for SQL requests. About three-quarters of threads originated outside the analytics team, while 85% received a first response within a minute. Grab reports that this shift reduced analysts’ mechanical workload, allowing more focus on self-service workflows and deeper analysis.