Ajay Agrawal|Java · Kafka · Search Architect
AGENTIC-READY COMMERCE SEARCH & MIGRATION INSTRUMENT

Anvi Search: Next-Generation Enterprise Search & RAG Architecture

Anvi Search is an enterprise commerce search platform engineered as a drop-in Oracle Endeca migration instrument and the world's first Agentic-ready hybrid retrieval system — combining Model Context Protocol (MCP) tool contracts, Graph-RAG relational knowledge traversal, dense vector retrieval, explainable merchandising, and 100% private in-VPC data isolation.

1. Executive Overview & The Migration Dilemma

Oracle Endeca (MDEX) is end-of-life, as are sustaining-support ATG Commerce and WebSphere Commerce search installations. Thousands of the world's largest retailers and B2B distributors still run multi-billion-dollar revenue streams through them. These migrations historically stall for two reasons that have nothing to do with basic search indexing:

  1. Inability to prove empirical relevance on the retailer's real catalog: Vendor ranking decisions are traditionally made on synthetic vendor demos rather than rigorous scientific measurements, causing enterprise architects to defer switching.
  2. Decades of merchandising rules, triggers, and dimension hierarchies would be destroyed: Thousands of category triggers, dynamic experience cartridges, dimension precedence rules, boost/bury actions, and pinned slots represent millions in institutional investment. Merchandisers correctly view a traditional platform rewrite as pure operational risk.

Anvi Search is built as an Endeca Migration Instrument. It preserves the Endeca conceptual model so business merchandisers do not retrain, provides an explainable score trace for every ranking decision, and ships the offline evaluation harness to measure relevance against your actual catalog before cutover.

Anvi is derived from the Sanskrit अन्वेषण (anveshaṇa) — meaning search, seeking, deep investigation, and systematic inquiry.

2. Agentic AI, MCP Tool Contracts & Advanced RAG

Search in 2026 is no longer just a search box for human shoppers; it is the authoritative sensory and retrieval layer for autonomous AI agents. Anvi Search is built from the ground up with native Agentic AI protocols:

Native Model Context Protocol (MCP) Server

Exposes GET /mcp/tools allowing autonomous AI agents (Claude, OpenAI, Gemini) to inspect catalog schema, query faceted attributes, and execute parametric lookups with dynamic JSON Schema generation.

Agent Readiness & Hallucination Shield

Via GET /readiness, Anvi audits product attribute completeness and generates cryptographic citation proofs, guaranteeing LLM agent recommendations are 100% grounded in verified SKU facts.

Graph-RAG Relational Knowledge Traversal

Traverses complex product compatibility matrices, variant trees, and bundle graphs at query time to answer relational questions (e.g. "which HEPA filter fits my 2021 air purifier?").

Corrective RAG (CRAG) & Confidence Filtering

Automatically scores retrieval confidence across hybrid branches, pruning noisy or low-relevance catalog snippets before feeding generation contexts to optimize agent token budgets.

3. The 10-Stage Request Lifecycle

Every search query executes through an explicit, deterministically ordered 10-stage lifecycle. In Anvi Search, order is the design — each stage may only shape what follows it:

01
Request Ingestion & Context Binding (0.2 ms)

Extracts q, nav_state, profile (audience segment), site, and page_type. Strips spoofable tenant headers and validates cryptographic key scopes.

02
Parse & Query Normalization (0.4 ms)

Unicode case-folding, accent normalization, alphanumeric dimension splitting (e.g. 12v, 100w, 1/2in), and token extraction.

03
Query-Side Synonyms Expansion (0.6 ms)

Merchandiser-editable equivalence and one-way sets. Executed before rules so rules evaluate against expanded queries rather than raw strings.

04
Landing Page & Trigger Resolution (1.1 ms)

Depth-before-breadth candidate walk. The single winning page defines the baseline layout, hero banners, and promotional cartridges.

05
Business Merchandising Rules Engine (0.9 ms)

Layers dynamic boosts, buries, pins, and redirects on top of the winning page's query plan.

06
Dual-Branch Concurrent Retrieval (9.7 ms)

Branch A (Lexical): Lucene/Solr edismax inverted index with configurable match modes (all_partial).
Branch B (Vector k-NN): 384-dimensional quantized e5-small-v2 dense semantic embeddings.

07
Application-Layer RRF Hybrid Fusion (1.8 ms)

Reciprocal Rank Fusion (Lexical 1.0 / Vector 0.4) executed in the application tier to ensure Endeca sort stacks (Price ASC, Margin) are never dropped.

08
Opt-In Cross-Encoder Neural Reranking (115 ms)

Deep cross-attention via ms-marco-MiniLM-L6-v2 (+0.035 nDCG boost). Merchandiser pins are strictly preserved and never reordered.

09
Guided Navigation & Vocabulary Speller (3.5 ms)

Computes multi-select facet counts. Speller resolves against index vocabulary first, firing only when primary results are empty.

10
Layout Presentation & Diagnostic Egress (0.3 ms)

Emits structured JSON with layout zones (Hero, Promo Rail, Product Grid, Facets, and per-stage latency audit metrics).

NEXT-GENERATION ENTERPRISE ARCHITECTURE

Complete 28-Subsystem Architecture Blueprint

Explore every component across Agentic AI, Model Context Protocol (MCP), Hybrid RAG, 10-Stage Pipeline, 3 Ingestion Lanes, and In-VPC Isolation.

✓ Production Shipping< 1 ms

Model Context Protocol (MCP) Server (GET /mcp/tools)

Standardized tool-calling contract allowing autonomous AI agents (Claude, OpenAI, Gemini) to query catalog facets, search products, and inspect inventory.

Industry-First: Native MCP Server built directly into the search engine core.
Endeca Parity: Not Supported in Endeca (Pre-AI Era Architecture)
✓ Production Shipping< 2 ms

Agent Readiness & Grounding Shield (GET /readiness)

Guarantees zero AI hallucination by evaluating catalog attribute completeness and generating verified citation proofs for LLM outputs.

Hallucination Shield: Mathematically guarantees LLM agent answers are grounded in real SKU facts.
Endeca Parity: Not Supported in Endeca
✓ Production Shipping3.8 ms

Graph-RAG Relational Knowledge Traversal

Traverses product compatibility matrices, variant trees, and bundle graphs to resolve relational questions.

Graph-RAG: Connects parts, accessories, and compatibility graphs inside the search pipeline.
Endeca Parity: Custom ATG/Endeca Relational Cartridge Feeds
✓ Production Shipping1.5 ms

Corrective RAG (CRAG) & Confidence Filtering

Evaluates retrieval confidence across hybrid branches; automatically filters out low-relevance items before feeding generation contexts.

Self-evaluating retrieval that guards against low-quality RAG contexts.
Endeca Parity: Not Supported in Endeca
⚙ Roadmap2.1 ms

In-Session Real-Time Intent Drift Tracking

Dynamically updates user intent vectors within the active shopping session based on click paths without requiring persistent PII.

Real-time vector drift tracking that personalizes without privacy invasion.
Endeca Parity: Endeca Experience Manager Profile Rules (Static only)
⚙ Roadmap18.5 ms

Multimodal Vision-Language Search (SigLIP / CLIP)

Joint vision-language embedding space enabling visual similarity search combined with text attributes.

Unified vision + text semantic space.
Endeca Parity: Not Supported in Endeca
✓ Production Shipping0.4 ms

Query NLU & Token Normalization

Deep query normalization, Unicode case-folding, punctuation normalization, and token-level accent stripping.

Endeca Parity: Endeca Pipeline Tokenizer & Character Mapping
✓ Production Shipping0.6 ms

Query-Side Synonyms Engine

Merchandiser-editable one-way and two-way equivalence synonym expansion running before rules evaluate.

Endeca Parity: Endeca Thesaurus (Equivalence & One-Way Sets)
✓ Production Shipping1.2 ms

NLP Intent & Entity Decomposition

Deconstructs compound queries into structured attribute filters (colors, sizes, brands, price boundaries).

Endeca Parity: Endeca Automatic Category & Dimension Value Recognition
✓ Production Shipping0.8 ms (zero results only)

Vocabulary-Aware Speller & Did-You-Mean

Spelling suggestion that validates against index vocabulary first, firing only when strict matches fail.

Endeca Parity: Endeca Did You Mean (DYM) & Automatic Spell Correction (Aspell)
✓ Production Shipping9.7 ms

Lexical BM25 / Edismax Engine

High-throughput Lucene/Solr inverted index retrieval with configurable multi-match modes.

Endeca Parity: Endeca Dgraph Text Search Engine (MatchMode)
✓ Production Shipping14.2 ms

Dense Vector Semantic Search (kNN)

384-dimensional dense semantic embedding retrieval powered by quantized e5-small-v2 ONNX models.

Endeca Parity: Not Supported in Endeca (Custom Extension)
✓ Production Shipping16.5 ms

Cross-Lingual Retrieval Engine

Retrieves English product catalog records from queries submitted in Spanish, French, or other languages.

Endeca Parity: Not Supported in Endeca (Required Separate Dgraph per language)
✓ Production Shipping1.8 ms

Application-Layer RRF Hybrid Fusion

Reciprocal Rank Fusion in the application tier (Lexical 1.0 / Vector 0.4) preserving sort stacks.

Endeca Parity: Not Supported in Endeca (Endeca only supported Stratified Ranking)
✓ Production Shipping115 ms (Opt-in)

Cross-Encoder Neural Reranker (ms-marco-MiniLM-L6-v2)

Deep sequence-pair cross-attention reranker providing +0.035 nDCG boost.

Endeca Parity: Not Supported in Endeca
⚙ Roadmap8.0 ms

Learning to Rank (LTR / Click-Feedback)

Machine learning ranking (LambdaMART) trained on verified live click, add-to-cart, and purchase streams.

Endeca Parity: Not Supported in Endeca
✓ Production Shipping1.1 ms

Landing Page & Trigger Resolution (Depth > Breadth)

Deterministic page and cartridge resolution walking the navigation state hierarchy.

Endeca Parity: Endeca Experience Manager (Page / Navigation State Triggers)
✓ Production Shipping0.9 ms

Merchandising Rules Engine (Boost, Bury, Pin)

Layered business rules modifying product elevations, score multipliers, redirects, and banner slots.

Endeca Parity: Endeca Business Rules & Boost/Bury Actions
✓ Production ShippingWorkbench UI

The Resolution Inspector (Visual Debugger)

Interactive diagnostic workbench answering Why is this page showing? with complete candidate traces.

Endeca Parity: Not Available in Endeca (Endeca required reviewing XML EAC logs)
✓ Production Shipping3.5 ms

Guided Navigation & Dynamic Faceting

Multi-select facet counts, hierarchical dimension trees, and dynamic facet reordering.

Endeca Parity: Endeca Guided Navigation & Dynamic Dimension Refining
✓ Production Shipping4.2 ms

Negotiable Retrieval on Zero Results

Analyzes binding constraints when queries return 0 results and suggests relaxed alternatives with live counts.

Endeca Parity: Not Supported in Endeca
✓ Production Shipping0.4 ms

Sub-Millisecond Typeahead & Autocomplete

Ultra-fast prefix search honoring tenant, entitlements, and merchandiser exclusions.

Endeca Parity: Endeca Dimension Value Search & Typeahead
✓ Production ShippingPipeline Sync

Lane A: Batch Catalog Sync & Blue-Green Reindex

12-step full catalog reindex with validated configsets, monotonic guards, and node parity validation.

Endeca Parity: Endeca Baseline Index Pipeline (Forge / Dgidx / EAC Baseline Update)
✓ Production Shipping< 20 ms CDC

Lane B: Real-Time Price & Inventory Streaming (CDC)

Microsecond price, stock, and status updates via high-watermark PostgreSQL streaming.

Endeca Parity: Endeca Partial Updates (Delta Feed / EAC Partial Index Update)
✓ Production ShippingAsync Graph Feed

Lane C: Dynamic Graph & Product Relations

Calculates and updates product bundles, cross-sells, and compatible accessories.

Endeca Parity: Custom ATG/Endeca Relational Cartridge Feeds
✓ Production ShippingRealtime Logs

Query Health & Fallback Analytics

Monitors search fallback rates, zero-result frequency, and shallow rerank pool warnings.

Endeca Parity: Endeca Search Analytics & Log Server
✓ Production ShippingAtomic TX

Atomic Append-Only Audit Trail

Complete, immutable record of every merchandising change, rule edit, and API key action.

Endeca Parity: Not Supported in Endeca (Endeca overwritten XML without version history)
✓ Production ShippingZero Egress

100% In-VPC Isolation & Data Protection

Runs entirely within your cloud VPC or bare-metal environment with zero external telemetry egress.

Endeca Parity: Endeca On-Premise Deployment Model
Production ShippingLatency Budget: < 1 ms

Model Context Protocol (MCP) Server (GET /mcp/tools)

Standardized tool-calling contract allowing autonomous AI agents (Claude, OpenAI, Gemini) to query catalog facets, search products, and inspect inventory.

Innovation Breakthrough: Industry-First: Native MCP Server built directly into the search engine core.

Technical Execution & Design Guarantees

  • Dynamically generates tool parameters and JSON Schema directly from the active catalog attribute registry.
  • Strict tenant and credential encapsulation preventing agent token leakage.
  • Allows multi-agent shopping assistants to perform complex facet filtering without hardcoded assumptions.
Oracle Endeca Drop-in Equivalent
Not Supported in Endeca (Pre-AI Era Architecture)
Enables drop-in migration without retraining business merchandisers or restructuring product catalog feeds.
VISUAL SEARCH PIPELINE ARCHITECTURE

Interactive End-to-End Query Flow

Click on any architectural block to inspect its latency budget, data inputs/outputs, and underlying technology stack.

⏱ 0.1ms
🛒
1. Query IntakeStorefront / MCP Agent
⏱ 1.2ms
🧠
2. NLU & Intent ParserTokenization, Entities & Synonyms
⬇ (Parallel Dual-Branch Execution)
⏱ 3.5ms
BRANCH A (SPARSE)
📖
3A. Lexical BM25 CoreExact SKU & Keyword Match
⚡ CONCURRENT
⏱ 4.8ms
BRANCH B (DENSE)
🔮
3B. Dense Vector k-NNSemantic Similarity Graph
⬇ (Reciprocal Rank Fusion Merge)
⏱ 0.6ms
🔄
4. Reciprocal Rank Fusion (RRF)RRF(d) = Σ w / (k + rank) · Calibrated Fusion
⏱ 2.4ms
🎯
5. Neural RerankerCross-Encoder Top-30 Re-scoring
⏱ 0.8ms
⚖️
6. Merchandiser RulesBoost, Bury, Pin & Cartridges
⏱ 1.1ms
🚀
7. Facets & DeliveryMulti-Select Disjunctive Facets < 10ms Total
STAGE INSPECTOR

4. Reciprocal Rank Fusion

Score-Free Ranked Merge
Latency Budget: 0.6ms

Merges lexical and vector candidate lists using calibrated RRF weighting (Lexical 1.0, Vector 0.4, k=60), solving vocabulary mismatch.

📥 Inputs & Triggers:
  • BM25 Candidates
  • Vector k-NN Candidates
📤 Outputs & Artifacts:
  • Unified Top 60 Hybrid Candidates
Engine Technology:
Application-Layer RRFMonotonic Rank Sorter
DATA INGESTION ARCHITECTURE

Three-Lane Multi-Speed Catalog Ingestion

How Anvi Search ingests high-scale catalog data: combining full-fidelity offline batch reindexing with sub-20ms streaming price/stock CDC.

Lane AHourly

Batch Blue-Green Reindex

Hourly / Nightly Baseline

Lane BContinuous

Real-Time CDC Streaming

Continuous (Sub-20ms Freshness)

Lane CGraph-RAG

Dynamic Relations Graph

Graph-RAG Periodic Linker

Lane B: Real-Time CDC Streaming

Continuous (Sub-20ms Freshness)
PRODUCTION TESTED

Captures row-level database changes (price modifications, stock level drops, flash sale flags) and streams them via CDC directly into an in-memory document overlay.

1
1. Transactional DB Change (Postgres / Oracle / Mongo CDC)
2
2. Kafka / Debezium Stream Event Router
3
3. High-Speed In-Memory Price/Stock Cache Mutation
4
4. Query-Time Document Attribute Overlay (< 20ms total)
Peak Throughput:10,000 events / sec
Propagation Latency:< 20ms end-to-end
Core Isolation:In-Memory Monotonic Watermark
Rollback Contract:Kafka offset rewind
MIGRATION & PARITY BLUEPRINT

Oracle Endeca vs. Anvi Search Architecture

How Anvi Search acts as an empirical migration instrument — replacing legacy MDEX components while preserving 100% of your business rules and storefront contracts.

LEGACY ARCHITECTURE

Oracle Endeca (MDEX 11.3)

01
Dgraph Engine (C++ In-Memory Index)

Single-threaded query processing bound to proprietary MDEX binary. Vulnerable to memory fragmentation.

02
Forge & Dgidx Pipeline

Brittle XML configuration files running multi-hour monolithic batch crawls.

03
Experience Manager (EAC Cartridges)

Proprietary XML rule engine for boost/bury, slotting, and dimension precedence.

04
Navigation State Query Protocol (N, Ne, Ntt, Ntk)

Proprietary query parameters embedded in ATG JSP / React storefront form handlers.

⚡ ZERO REWRITE TRANSLATOR
MODERN IN-VPC REPLACEMENT

Anvi Search Platform

01
Stateless In-VPC Hybrid Core

Concurrent multi-threaded C++ / Rust core executing dual BM25 + Vector HNSW retrieval.

02
3-Lane Ingestion (Batch + Sub-20ms CDC)

Continuous event streaming via Kafka / Debezium with in-memory delta overlays.

03
Dynamic Cartridge Engine + Resolution Inspector

1:1 drop-in XML import with real-time audit tracing and explainable rule collision visualizer.

04
Drop-In MDEX Query Protocol Translator

Transparently accepts legacy Endeca query parameters and maps them to hybrid plans.

MIGRATION PARITY INSPECTOR: COMPONENT 01

Dgraph Engine Stateless In-VPC Hybrid Core

⚠️ Legacy Endeca Bottleneck:

Unpredictable latency spikes under heavy concurrent faceting.

✓ Anvi Search Advantage:

Sub-10ms predictable p99 latency across millions of SKUs with horizontal auto-scaling.

AGENTIC AI & MCP PROTOCOL ARCHITECTURE

Model Context Protocol (MCP) Grounding Pipeline

How autonomous AI shopping agents interact with Anvi Search via standard MCP tool contracts — ensuring 100% factual accuracy and zero telemetry egress.

010.2ms

Dynamic Tool Contract Discovery

AIAnvi
021.1ms

Agent Tool Execution Call

AIAnvi
034.5ms

Hallucination Shield & In-VPC Verification

AnviPrivate
040.3ms

Grounded Citation Payload Delivery

AnviAI
STEP 01 PROTOCOL INSPECTOR

Dynamic Tool Contract Discovery

AI Shopping Agent (Claude / Cursor)Anvi In-VPC MCP Gateway

The AI Agent queries the MCP gateway to discover active catalog dimensions, available tools, and parameter schemas dynamically without hardcoded prompting.

GET /mcp/tools -> Tool RegistryJSON PAYLOAD
{
  "tools": [{
    "name": "catalog_search",
    "description": "Hybrid vector + BM25 catalog search with parametric facet filtering.",
    "parameters": {
      "query": { "type": "string" },
      "filters": { "type": "object" }
    }
  }]
}

4. The Three Data Ingestion Lanes

Commerce data changes at three fundamentally different cadences. Rather than forcing all updates through a single bottleneck, Anvi Search separates ingestion into three dedicated lanes:

Lane A: Full Catalog Reindex (Batch)

12-step automated blue/green collection reindexing. Features validated configset pushes, monotonic timestamp guards, and multi-node parity assertions to guarantee deterministic tie-breaking.

Lane B: Price & Stock CDC (Real-Time)

Sub-20ms streaming price and stock availability updates via POST /price/{domain} with PostgreSQL high-watermarks. Maintains live stock accuracy during flash sales without reindexing.

Lane C: Relational & Graph Feeds (Async)

Updates dynamic cross-sell matrices, product bundles, and compatible accessories via POST /relations on the zero boundary for ERP and PIM synchronization.

Node Parity & Determinism Guard

Guarantees that identical documents across replica nodes receive identical ranking scores and deterministic tie-breaking, preventing pagination jump defects.

5. Merchandiser Workbench & The Resolution Inspector

A major point of friction in search migrations is merchandiser enablement. Anvi Search provides 8 purpose-built workbench screens designed around the mental model of enterprise commerce operators:

Workbench ToolCapabilities & Operational Benefit
The Resolution InspectorAnswers "Why is this page showing?" for any navigation state. Evaluates every candidate in the category hierarchy and displays distinct verdicts (Won, Never Published/Draft, Expired Schedule, Not Found).
Visual Category & Tree EditorHierarchical navigation management with drag-and-drop category restructuring and dimension inheritance.
Landing Page & Cartridge BuilderVisual assembly of promotional banners, rich text, carousels, and product elevation widgets.
Unified Segmentation EngineA/B test variants and personalization use the same depth-before-breadth hierarchy — zero separate segment databases to sync.
Atomic Append-Only Audit TrailEvery rule modification and synonym change is committed atomically within the database transaction, providing tamper-evident governance logs.

6. Empirical Benchmarks (Amazon ESCI Dataset)

Relevance claims should be established through rigorous empirical measurement, not vendor assertions. Anvi Search was evaluated against 4,998 real commerce products and 5,482 human relevance judgements across 191 queries from the Amazon ESCI benchmark (graded Exact, Substitute, Complement, Irrelevant):

Retrieval ConfigurationnDCG@10 ScoreRecall@50 (Grade ≥ 2)Empirical Characteristics
Pure Lexical (BM25 / Edismax)0.5270.494High precision on exact SKUs and model numbers; produces 5 zero-result queries on vocabulary gaps.
Pure Vector (Dense Embeddings)0.4920.575Captures broad conceptual intent but loses to lexical on exact alphanumeric part queries.
Anvi Hybrid Fusion (RRF)0.5370.645Superior overall relevance: 0 zero-result queries (100% coverage) while preserving exact match precision.
Anvi Hybrid + Neural Reranker0.5690.648+0.035 nDCG lift via cross-encoder sequence-pair attention; strictly preserves merchandiser pins.
Note: Retracted an earlier synthetic claim that vector beat lexical by +0.206. On real human judgements, vector alone loses to lexical; hybrid fusion wins because the branches fail on completely different query classes.

7. Oracle Endeca Drop-in Feature Parity Matrix

Anvi Search maps directly to Endeca concepts, allowing automated migration of existing rule pipelines and EAC configurations:

Endeca Concept (MDEX / EAC)Anvi Search ArchitectureMigration & Operational Impact
Experience Manager Pages & TriggersLanding Pages & Nav State Triggers1:1 automated import; zero retraining required for business merchandisers.
Dimension Precedence RulesHierarchical Dimension Trees & Attribute RegistryPreserves category drilldown logic without writing custom middleware.
Match Modes (MatchAll, MatchAny, Partial)Configurable Edismax Query ParsersDrop-in query syntax compatibility for existing commerce frontends.
Merchandising Cartridges (Banners, Carousels)Layout Zone WidgetsMaintains identical CMS slotting and promotional payload structures.
Stratified Ranking / Boost & BuryMulti-Stage Hybrid RRF + Score MultipliersReplaces rigid integer strata with explainable mathematical multipliers.
Forge / Dgidx Baseline PipelinesLane A Batch Ingestion with Parity CheckReplaces complex proprietary pipeline scripts with declarative JSON/Postgres feeds.
EAC Delta UpdatesLane B Real-Time CDC StreamingSub-20ms price and stock updates instead of hourly batch deltas.
In-Memory Dgraph ServerHigh-Throughput Lucene/Solr + Vector Core100% In-VPC deployment with modern multi-core parallelism and horizontal scalability.

8. Security, Privacy & 100% In-VPC Isolation

Unlike third-party multi-tenant SaaS search providers that ingest your proprietary catalog and customer search events into external cloud environments, Anvi Search runs 100% inside your enterprise perimeter:

  • Zero External Data Egress: No search queries, customer identifiers, or catalog records ever leave your private VPC or bare-metal infrastructure.
  • Privacy-by-Design: Zero personal data processed by default. Behavioural analytics requires explicit configuration with keyed HMAC pseudonymization.
  • Regulatory Compliance: Fully compliant with GDPR, CCPA, and India DPDP Act. Subject access and erasure (DSAR) are built-in native operations.
  • AI Agent & MCP Tool Readiness: Implements standard Model Context Protocol (GET /mcp/tools) for seamless, secure integration with enterprise AI agents.