Skip to main content

FrootAI — AmpliFAI your AI Ecosystem Get Started

All Solution Plays

Play 26

Semantic Search Engine

Medium Designed

Hybrid search with reranking, personalization, and answer generation.

A complete search experience powered by Azure AI Search and LLMs. Combines full-text search (BM25), vector search (embeddings), and hybrid fusion with semantic reranking. Query expansion uses GPT to generate alternative phrasings. Personalization layer adapts results based on user history. Answer generation synthesizes a direct answer from top results with citations.

Architecture Pattern

Hybrid search: BM25 + vector + reranking, query expansion, answer generation

Azure Services

Azure AI SearchAzure OpenAI (gpt-4o)Blob StorageContainer Apps

DevKit (.github Agentic OS)

  • agent.md — root orchestrator with builder→reviewer→tuner handoffs
  • 3 agents — Search Builder (gpt-4o), Reviewer (gpt-4o-mini), Tuner (gpt-4o-mini)
  • 3 skills — deploy (107 lines), evaluate (105 lines), tune (103 lines)
  • 4 prompts — /deploy, /test, /review, /evaluate with agent routing
  • .vscode/mcp.json — FrootAI MCP with AI Search + OpenAI inputs + envFile

TuneKit (AI Config)

  • config/search.json — hybrid weights (60/40), reranker model, top-k
  • config/openai.json — query expansion + answer gen prompts
  • config/guardrails.json — content filtering, PII in search results
  • evaluation/eval.py — NDCG@10 >0.75, Answer accuracy >85%

Tuning Parameters

Hybrid weights (full-text vs vector)Reranker model selectionPersonalization featuresAnswer generation styleQuery expansion depthTop-k results (5→20)

Machine evidence

FrootAI evidence lifecycle

This is an internal evidence maturity label, not third-party certification, accreditation, legal compliance, or a production guarantee. Missing or expired evidence demotes automatically; catalog claims cannot promote a play.

Designed
designed
build verified
evaluation verified

This play currently has design evidence only. A runnable scenario, endpoint evaluation, and build receipts are the next contiguous gates.

Loading architecture and cost model…

Repo Intelligence

v1

A no-clone, revision-pinned map for agents and humans. Observed evidence is separated from inferred flow so the output stays useful without pretending to be a full call graph.

Indexing bounded repository evidence…