Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.vh3.ai/llms.txt

Use this file to discover all available pages before exploring further.

​
Semantic Search Guide

VH3 AI’s search is hybrid — it combines semantic search (understanding meaning) with BM25 keyword search (matching exact terms). This means you get relevant results whether you describe a problem conceptually or use specific technical terminology.

​
How it works

When you submit a search query, the platform:
  1. Embeds your query into a vector representation using the same model that embedded all job text at ingestion time
  2. Runs BM25 keyword matching in parallel for exact term hits
  3. Fuses the results using reciprocal rank fusion to produce a single ranked list
This means “boiler intermittent fault” will find jobs recorded as “heating system cutting out,” “CH unit tripping intermittently,” or “combi boiler loss of flame.”

​
Tips for effective queries

​
Describe the problem, not the solution

“air conditioning not cooling despite running”

​
Use natural language

The search engine understands context. Write queries the way you’d describe the problem to a colleague.

​
Combine with filters

For precision, combine semantic queries with structured filters: 
{
  "companyId": "your-company-id",
  "query": "water leak from ceiling",
  "filters": {
    "startDate": "2024-01-01",
    "endDate": "2024-06-30",
    "siteKey": "site-tesco-manchester"
  }
}

​
Search types

EndpointSearches acrossBest for
/search/outcomesAI-enriched job outcomes”What happened?” / “How was it fixed?”
/search/intakeRaw fault descriptions”What was reported?”
/search/summary-sectionsGenerated report summariesFinding previous analysis

​
Use cases

  • Pre-visit preparation — search for similar past faults at a site before attending
  • Diagnostic support — find how similar problems were resolved historically
  • Trend detection — search for a fault pattern and see if it’s emerging across sites
  • Knowledge transfer — new engineers searching the collective experience of the team