AI Technology
Natural Language Search
Query your entire security and fleet dataset in plain English — no filters, no date pickers, no report configuration. Just ask the question and Akilapa Guardian answers instantly.
How it works
Natural Language Search converts plain-English questions into structured database queries and AI analysis, returning answers with supporting evidence.
Step 1
Query Interpretation
The NL engine parses the user's question to identify intent, entities (vehicles, drivers, dates, locations), and the type of response expected (data, map, chart, or narrative).
Step 2
Data Retrieval
The interpreted query is translated into optimised database queries across the platform's data lake — returning relevant records, events, and aggregated metrics.
Step 3
Response Generation
Results are formatted for the query type — a map for location questions, a table for list questions, a chart for trend questions, and a narrative paragraph for complex analysis.
Data inputs
- All platform data: trips, alerts, events, scores, incidents
- User's organisational context (their assets, drivers, sites)
- Query history for intent disambiguation
Outputs
Direct Answer
Plain-English response to the question
Supporting Evidence
Data table, map, or chart supporting the answer
Follow-up Suggestions
Related questions the user might want to ask next
Example insight
Query: 'Which drivers had more than 5 harsh braking events last week?' → Returned: 4 drivers. Chukwudi Okafor (12 events), Musa Ibrahim (8 events)... [table with full data]. Suggested: 'Show Chukwudi's worst trip this week'
See Natural Language Search on your fleet
Talk to our security architects about vehicles, fleets, churches, schools, and enterprise assets.