AI INFRASTRUCTURE FOR YOUR DATA

Build AI That Knows Your Data.

Connect your knowledge, create intelligent retrieval pipelines, and deploy grounded AI experiences from one cloud platform.

Start with your own documents. Move to APIs and connected data when you're ready.

truetales-pipeline/rag-orchestrator-v2
Demo Data / Illustrative Example
01. Ingestion Data SourcesActive Stream: 6 Source Connectors
PDF142 docs
DOCX68 files
TXT31 docs
Markdown89 docs
Web Content1,200 urls
API DataSync live
TRUE TALES DATA HUB

Multi-tenant ingestion layer with automatic schema extraction and metadata isolation.

PROCESS ENGINE
ParseNormalizeChunkEnrich
KNOWLEDGE INDEX

Hybrid vector index + sparse BM25 token index with cosine similarity partitions.

RETRIEVAL & MODEL

Cross-encoder reranking, context assembly & grounded model generation.

User Query“What are the cancellation terms for enterprise customers?”
Latency: 114ms
Retrieved Source Cards (Top 3 Context Slices)
Policy.pdfScore: 0.94
Section 4: Enterprise Contract Termination
Enterprise TermsScore: 0.91
Article 7.2: Voluntary Cancellation Clause
Contract GuideScore: 0.88
Notice Period & SLA Reconciliation
Grounded Response Generated

“Enterprise customers may cancel according to the notice period defined in their agreement. Cancellation requests must be submitted in writing with a minimum of 30 days notice prior to the end of the current billing cycle.”

Sources Cited:Policy.pdf — Section 4Enterprise Terms — CancellationContract Guide — Notice Period
CORE ARCHITECTURE

Everything Between Your Data and Your AI.

A comprehensive, developer-first infrastructure stack designed to turn unorganized corporate files into verified, low-latency AI intelligence.

Ingestion & ETL

Data Hub

Bring your knowledge into one AI-ready layer.

Unified ingestion pipelines for enterprise documents, crawled web knowledge, structured files, and authenticated private APIs.

  • PDF, DOCX, TXT, MD, CSV
  • Automatic text normalization
  • Streaming metadata tags
  • Zero data retention leakage
Explore Data Hub
Indexing & Vectors

Knowledge Indexes

Transform raw content into searchable knowledge.

Deconstruct complex documents into semantic chunks with high-dimensional vector embeddings and token-level sparse keyword indexes.

  • Configurable chunk overlap
  • Dense + Sparse hybrid matrix
  • Metadata filter partitions
  • Real-time index updates
Explore Knowledge Indexes
Hybrid Retrieval

RAG Engine

Retrieve the right context before your model answers.

Precision hybrid retrieval orchestrator combining cosine similarity vectors, BM25 keyword matching, and cross-encoder neural reranking.

  • Dense/sparse weighting
  • Top-K and similarity cutoffs
  • Context deduplication
  • Strict citation verification
Explore RAG Engine
Evaluation Suite

AI Playground

Test retrieval and grounded answers before writing integration code.

Full developer testbed to query indexes in real time, inspect chunk relevancy scores, evaluate latency, and tune parameters.

  • Side-by-side prompt testing
  • Relevance score heatmaps
  • Latency telemetry (ms)
  • One-click API export
Explore AI Playground
Agentic Workflows

Agent Builder

Create AI experiences that use your private knowledge.

Deploy goal-driven assistants with direct bindings to enterprise knowledge indexes, role instructions, and enforced citation guarantees.

  • Enforced source citations
  • Multi-index scoping
  • Deterministic guardrails
  • Deployable via REST & Webhook
Explore Agent Builder
Knowledge Topology

Context Graph

Explore how entities, topics and knowledge connect.

Interactive topological graph mapping cross-document dependencies, policy hierarchies, entity relationships, and knowledge clusters.

  • Entity-relationship extraction
  • Cross-document lineage
  • Interactive node visualizer
  • Configured entity schemas
Explore Context Graph
DATA HUB

Connect the Knowledge Your AI Is Missing.

Your AI cannot answer what it cannot read. TrueTales Data Hub standardizes fragmented enterprise documents into an AI-ready layer, normalizing unstructured text while maintaining document lineage.

Supported Formats:PDF, DOCX, TXT, Markdown, CSV
Connectors:Postgres, S3, Snowflake (Coming Soon)
Data Sources4 active repos
Product Documentation
124 documentsPDF / MD
Support Knowledge
82 documentsDOCX / TXT
Company Policies
36 documentsPDF
Technical Manuals
211 documentsMD / PDF
Enterprise encryption in transit & at restDemo Data
KNOWLEDGE INDEXES

Turn Raw Data Into Retrieval-Ready Knowledge.

Raw text is noisy. TrueTales processes, chunks, enriches with metadata, and generates dual vector-sparse indexes so queries match the exact semantic context.

01
Document
Raw file ingestion
02
Parsing
Text & tables cleaned
03
Chunking
Recursive semantic split
04
Metadata
Source, date & tags
05
Embeddings
1536-dim vector tensor
06
Vector Index
HNSW hybrid search
Sample Production Index

Product KnowledgeDemo Data

Status: Ready
453
Documents
8,420
Chunks
2 min
Last Updated
Dimensions: 1536 (Cosine)Inspect Index Settings
RAG ENGINE

Ground Every Answer in Relevant Context.

Control how your application finds, ranks and sends knowledge to your model.

USER QUERY
Input Intent
QUERY PROCESSING
Hypothetical Docs & HyDE
HYBRID / VECTOR RETRIEVAL
Dense HNSW + BM25
OPTIONAL RERANKING
Cross-Encoder Scorer
CONTEXT ASSEMBLY
Token Window Packing
MODEL
Configured LLM
RESPONSE + SOURCES
Verified Citations

Configurable Retrieval Controls

Fine-tune the mathematical precision of your retrieval pipeline. Balance recall vs precision, enforce strict relevance filters, and eliminate hallucinations with deterministic rerankers.

Top K (Context Slices)5 chunks
Similarity Cutoff Threshold0.78
Cross-Encoder Neural Reranker
Retrieval Strategy
View RAG Engine Documentation
Active RAG Execution PlanStatus: Validated
retrieval_mode: "hybrid"
dense_embedding: "text-embedding-3-small (1536)"
sparse_index: "bm25_v2 (k1=1.2, b=0.75)"
rerank_model: "bge-reranker-large"
top_k_candidates: 5
min_similarity_score: 0.78
metadata_filters: { tenant_id: "prod_ent", status: "published" }
citation_enforcement: "STRICT_REQUIRED"
Note: Configure real-time pipelines programmatically via Python/TypeScript SDK or REST API.
AI PLAYGROUND

Test Your AI Before You Ship It.

Validate retrieval recall, preview assembled model context, and verify citations directly in the browser.

Playground Session/Support Knowledge
Illustrative ExampleOpen Full Playground
Session Config
Customer Support AI
Support Knowledge
Configured Model (LLM-Grounded)
Enabled
Citations enforced with source traceability.
U
Can customers change their subscription during an active billing period?
AI

Yes. Customers may upgrade or modify their subscription tier at any point during an active billing cycle [Billing Policy §2.1]. Upgrades take effect immediately with prorated billing adjustments applied automatically [Subscription Guide].

100% grounded in company knowledge base
Retrieved Context
Source 1: Billing PolicyScore: 0.89

“Mid-cycle tier modifications are prorated based on remaining calendar days...”

Source 2: Subscription GuideScore: 0.82

“Customers receive instant feature activation upon confirmation of tier upgrade...”

Latency:142 ms
Retrieved Chunks:4 chunks
Token Usage:328 tokens
CONTEXT GRAPH

See How Your Knowledge Connects.

Explore relationships between entities, topics, documents and concepts across your knowledge base.

Interactive Knowledge MapClick any node to inspect extracted relationships
Extracted and configured relationships from multi-document ingestion.
Node Details

Enterprise Plan

Connected Context

Top-level enterprise subscription tier containing custom licensing, billing, and SLA parameters.

Related Documents
Enterprise_Agreement_2026.pdfIndexed
Pricing_Schedule_Master.docxIndexed
Related Concepts
Custom Billing CyclesDedicated InfrastructureSLA Guarantees
Connected Entities
Contract ID: ENT-994
Billing Org: Acme Global
Assigned Rep: Level 4
AGENT BUILDER

Build AI Experiences Around Your Knowledge.

Create specialized AI agents equipped with role instructions and bound to exact private knowledge indexes. Every response is strictly grounded with verifiable source citations.

Grounded Assistants, Not Black Boxes:TrueTales agents are engineered to retrieve and cite organization knowledge with deterministic constraints, not to make unchecked autonomous business decisions.
Target Workflows:
Customer SupportInternal KnowledgeProduct AssistantResearch AssistantDocumentation Assistant
CREATE AGENT
Agent Studio
Support Assistant
Answer questions using company support documentation. If an answer cannot be grounded directly in the provided knowledge indexes, politely decline and provide the support link.
Support Index82 docs
Product Index124 docs
Retrieval:Enabled
Citations:Required
DEVELOPER API

From Playground to Production With One API.

Integrate grounded intelligence into any backend, microservice, or web application with uniform low-latency REST endpoints.

POST/v1/query
curl -X POST https://api.truetalesworld.org/v1/query \
  -H "Authorization: Bearer tt_live_sec_89df20a" \
  -H "Content-Type: application/json" \
  -d '{
    "index": "product-knowledge",
    "query": "What is our enterprise cancellation policy?",
    "top_k": 5,
    "rerank": true
  }'
Authentication: Scoped Bearer TokensConceptual Example
Response 200 OKapplication/json
{
  "answer": "Enterprise customers may cancel according to the notice period defined in their agreement. Notice must be submitted in writing at least 30 days prior...",
  "sources": [
    { "title": "Policy.pdf", "section": "4", "score": 0.94 },
    { "title": "Enterprise Terms", "section": "Cancellation", "score": 0.91 }
  ],
  "usage": {
    "prompt_tokens": 184,
    "completion_tokens": 42,
    "latency_ms": 114
  }
}
SDK-Ready
Python & TypeScript
Request Logs
Trace token & latency
Rate Limits
10,000 req/min burst
API Keys
Fine-grained RBAC scopes
PIPELINE WORKFLOW

From Raw Data to Grounded AI in Five Steps.

A linear, dependable path from unindexed files to production-ready enterprise intelligence.

01

CONNECT

Upload or connect knowledge from documents, web crawls, or private endpoints.

02

INDEX

Process and organize content with automated semantic chunking, metadata, and embeddings.

03

CONFIGURE

Choose retrieval settings, Top-K thresholds, reranking behavior, and model parameters.

04

TEST

Evaluate questions inside the AI Playground, verifying source relevance and latency.

05

DEPLOY

Integrate through a unified REST API or deployed agent experience into your production stack.

ENTERPRISE SOLUTIONS

Engineered for Grounded Business Applications.

Proven retrieval architectures configured for real-world enterprise operations.

Customer Support AI

Build support assistants grounded in product manuals, SLAs, and warranty policies to resolve customer inquiries with verifiable citations.

Explore Solution Architecture

Internal Knowledge Search

Help engineering, ops, and legal teams instantly search company knowledge across hundreds of fragmented repositories.

Explore Solution Architecture

Documentation AI

Add intelligent semantic search and code Q&A to technical developer documentation with zero hallucinations.

Explore Solution Architecture

Research & Synthesis

Analyze and synthesize massive collections of technical whitepapers, contract revisions, and compliance guidelines.

Explore Solution Architecture

AI Applications Layer

Give existing SaaS platforms, web apps, and enterprise backends an integrated retrieval layer over proprietary customer data.

Explore Solution Architecture
DEVELOPER EXPERIENCE

Built for the Path From Prototype to Production.

A developer platform crafted for engineers who need observability, clean APIs, and predictable infrastructure.

Projects

Isolate distinct applications, environments, and client tenants.

Data Sources

Connect documents, web scrapers, and streaming payload endpoints.

Knowledge Indexes

Fine-tune chunk sizes, overlap parameters, and dense embeddings.

AI Playground

Evaluate queries, latency, and context assembly in the browser.

API Keys

Generate granular read/write keys with per-endpoint scoping.

Request Logs

Inspect full execution traces, chunk relevance scores, and token logs.

Usage Analytics

Track query volumes, p95 latency, and model tokens in real time.

Dev / Prod Envs

Promote validated indexes from sandbox to high-concurrency production.

OBSERVABILITY & TRACING

See What Your AI Retrieved.

TrueTales is built around total retrieval transparency rather than black-box AI.

Execution Trace: req_992f_live
Total: 114ms Tokens: 226 0 Errors
1. Query"What are the cancellation terms for enterprise customers?"
12ms
2. Hybrid RetrievalRetrieved 5 candidate chunks from Knowledge Index (Dense + BM25)
38ms
3. Cross-Encoder RerankingReranked to Top 3 most relevant slices (scores: 0.94, 0.91, 0.88)
24ms
4. Context AssemblyToken packing, deduplication & citation tags injected
6ms
5. Grounded ResponseSynthesized answer with verified citations (§4.0 & §7.2)
34ms
SECURITY & ARCHITECTURE

Your Data Should Stay Under Your Control.

TrueTales is built for enterprise engineering teams. We do not use your proprietary documents to train foundation models.

Workspace Isolation

Strict multi-tenant cryptographic logical boundaries preventing cross-account access.

Role-Based Access Control

Granular permissions for developers, workspace admins, and service accounts.

Secure Scoped API Keys

Revocable bearer tokens with fine-grained index-level and operation-level constraints.

Data Deletion Controls

Complete, irreversible deletion of documents, chunks, and vector embeddings upon request.

Encrypted Transport & Rest

Industry-standard TLS 1.3 in flight and AES-256 encryption at rest for all stored knowledge.

Audit-Friendly Logs

Comprehensive telemetry and immutable request traces for internal security reviews.

TRANSPARENT ARCHITECTURE PRICING

Scale From Early Prototype to High-Volume Cloud.

Metered on stored data, indexed chunks, and retrieval volume. No hidden artificial multipliers.

STARTER
Free Pilot

For developers testing AI on private knowledge.

Up to 250 MB Stored Data
10,000 Indexed Chunks
2,500 Queries / month
1 Workspace & 1 Index
Community Support
Start Building
PROPopular
$79/ month

For production applications and growing SaaS products.

5 GB Stored Data
250,000 Indexed Chunks
50,000 Queries / month
Hybrid Search + Cross-Encoder Reranker
Standard REST API & Webhooks
Email & Technical Support
Get Started
BUSINESS
$299/ month

For growing teams with complex knowledge hierarchies.

25 GB Stored Data
1,500,000 Indexed Chunks
250,000 Queries / month
Context Graph Entity Extraction
Multiple Environments (Dev/Prod)
Priority Technical Support
Get Started
ENTERPRISE
Custom

For advanced infrastructure, compliance, and custom SLAs.

Custom Storage & Chunk Volumes
Dedicated Cloud Retrieval Clusters
Custom Rate Limits & High Concurrency
VPC Peering & Dedicated Deployments
Named Technical Account Manager
99.95% Availability SLA
Contact Sales
FREQUENTLY ASKED QUESTIONS

Clear Answers to Technical Questions.

Precise technical details based strictly on our implemented cloud architecture.

TrueTales AI Cloud is a developer-focused AI infrastructure and RAG platform that enables companies to ingest internal documents, build searchable vector/keyword knowledge indexes, execute low-latency hybrid retrieval pipelines, and deploy verified, grounded AI experiences via APIs and agents.

Your Data Already Knows the Answer.
Give Your AI a Way to Find It.

Connect knowledge, configure retrieval and build grounded AI experiences with TrueTales AI Cloud.

Instant setup • Zero model training on customer data • REST API & SDKs