A product of IQVigilant™  ·  Natural Language & Agentic AI Engine

AI answers you can trust
at the lowest token cost.

IQVigilant Answers Agents is the natural-language and agentic AI engine inside IQVigilant. Six answering methods — from zero-LLM native parsing to hybrid agentic planners — give you safety, auditability, and the lowest AI token spend in production. Licensed as part of IQVigilant — see full licensing & pricing ↗.

The answering engine behind IQVigilant's
Natural Language module.

Every time a user types a question in an IQVigilant-powered app, this engine decides how to answer it — picking the safest, cheapest, and most appropriate method from six available paths.

🔒

Deterministic Safety

Native query & charts answers and human-vetted deterministic plans run without an LLM at execution time. Zero hallucinations, zero token cost.

🤖

Agentic Power

Workflow agents and direct LLM answering give you full reasoning power when you need it — with confidence scoring and audit trails.

Hybrid Intelligence

An LRM planner generates structured plans from natural language, then a deterministic executor runs them — best of both worlds.

🛡️

AI Vigilance & Governance

Tools discovered from your existing apps and APIs are fully manageable — enable or disable them, and control access permissions by user role. Enterprise safety, built in.

Six methods. Every answer, engineered.

Because one size doesn't fit all. A question like "show me all projects with their tasks" should be answered with zero-LLM native parsing — fast, deterministic, free. A question like "what's the strategic risk in our supply chain?" needs agentic reasoning. IQVigilant Answers Agents picks the right tool for every question, and lets you audit every decision.

Deterministic

Native query and charts answers

Answers questions directly from your database — no LLM required. Delivers data tables, numeric summaries, and charts (bar, line, pie, scatter, histogram, waterfall, heatmap, radar, and more) by mapping natural language to your data model.

Confidence: per-entity match percentage
Deterministic

Deterministic Plan

Predictable and repeatable. Enterprise-approved sequences of actions that run the same way every time. Plans can be inherited from agentic workflows — promote a successful agent run into a reviewed, hardcoded plan that executes with zero LLM cost at scale.

Confidence: 1.0 — human-vetted
Hybrid

Agentic Planner + Deterministic Executor

A reasoning LLM generates a structured plan from natural language asks, then the deterministic executor runs it. High-confidence plans are saved as candidates for promotion.

Confidence: LRM plan_confidence_score
Probabilistic

Workflow Agent

A full ReAct agent that leverages your existing agents, tools, and APIs. Give it access to your tool ecosystem — coded tools, NL-defined tools, and external APIs — for exploratory and power-user workflows.

Confidence: agent self-assessment
Probabilistic

Direct LLM

Sends the question straight to the configured LLM backend. Best for open-ended questions, summaries, and creative interpretation — can return formatted text, tables, lists, and structured explanations. Full audit trail preserved.

Confidence: configurable default (0.5)
Probabilistic

PandasAI

Uses the PandasAI library to answer data-statistics and chart questions from previously-retrieved DataFrames. Ideal for follow-up analytical queries.

Confidence: configurable default (0.9)

Method comparison

Method Family Consumes LLM tokens What the user sees
Native query & charts Deterministic None Data tables, summaries, charts
Deterministic Plan Deterministic None Agentic AI results, data, charts
Agentic + Deterministic Hybrid Low (planning only) Agentic AI results, data, charts
Workflow Agent Probabilistic Full Agentic AI results, text answers
Direct LLM Probabilistic Full Text, tables, formatted answers
PandasAI Probabilistic Full Text answers from data analysis

Safe AI Agents

Deterministic, agentic and hybrid modes, plus natural-language tool creation. The LLM acts as a safe translator into reviewed actions — no hallucinations, fewer tokens, lower cost of ownership.

Tools discovered from your existing apps and APIs become manageable — enable or disable them, and control access permissions by user role.


Learn more ↗

AI Querying, Charting & Analysis

Ask questions in plain language and get tables, charts and dashboards built from your data's metadata — across your own database, read-only external databases, and documents like Word, spreadsheets, PDFs and CSVs.


Learn more ↗

Ready to bring safe, auditable AI to your app?

Talk to us about licensing — start with a free 1-user license for personal use, or scale up at $99/month (8 users included, $10/user/month after). Or explore the full IQVigilant platform.

Talk to Us → IQVigilant Licensing ↗ Explore IQVigilant ↗