How we helped a healthcare startup increase patient onboarding by 180%

vervelo logo mobile
Analytics Agent — natural-language analytics
Data & Analytics Agent

Analytics Agent — Ask Your Data Anything

Ask questions in plain language and get charts, insights, and narratives — governed by your semantic layer, inside your own environment.

Answers, Not Just Dashboards

Most questions still funnel through a small data team and a backlog of dashboard requests. The Analytics Agent opens self-service analytics to everyone — safely — answering in plain language with charts and narratives, while staying governed by your metric definitions.

Plain English

Ask anything

No SQL or BI bottleneck required

Self-service

For every team

Answers without waiting on data

Chart + story

Every answer

Visuals with a written narrative

Governed

& cited

Uses approved metrics, shows its work

Analytics Agent Architecture

The Platform, Specialized for Analytics

The Analytics Agent runs on the same base Vervelo Agents Platform architecture as every other agent — here, the streaming runtime answers live queries, a durable runtime produces scheduled reports, and the connected tools are the warehouse, semantic-layer, and metric systems analytics depends on.

Analytics Agent Natural-language analytics — on the Vervelo Agents Platform DURABLE EXECUTION · SCHEDULED REPORTS SYNC EXECUTION · LIVE QUERIES Analyst Chat / Dashboard FastAPI Service Reporting Agent Analytics Agent Observability Warehouse / Lake (MCP) BI / Semantic Layer (MCP) Metrics / KPIs Definitions Query Store (Agents DB) Charts & Narratives Database Agent Definition Agent State Analyst Memory Query Sessions SQL Tools Warehouse MCP Semantic Vectors Prompt Development Flow ITERATE FEEDBACK LOOP Define Task & Success Criteria Metric defs, accuracy Draft Structure & Prompt SQL patterns, guardrails Manual Testing Sample questions & data Automated Evaluation Query-correctness eval suite Refine & Iterate Fix wrong SQL, bad joins Deploy & Monitor Track accuracy & adoption

Context Engineering

Engineered Context, Every Turn

Reliable agents aren't a single clever prompt — they're the product of context engineering: assembling exactly the right information into the model's context window on every turn. It's the discipline we build every Vervelo agent around. Here's what the Analytics Agent works from on each turn, and how we engineer it.

Analytics Agent context window

Assembled fresh on every turn

System Instructions

The analytics persona and rules — approved metric definitions, query guardrails, and never to invent numbers or ignore governance.

Persona & roleBehavioral rulesOutput format

Tool Definitions

Warehouse query, semantic-layer, and charting tools with strict schemas so queries are safe and results render correctly.

Name & descriptionI/O schemaUsage examples

Memory & User State

The analyst's context — prior questions, preferred metrics and formats, and standing definitions.

PreferencesPast decisionsStanding instructions

Conversation History

The analysis thread so far, so follow-up questions build on prior results and filters.

User turnsAssistant turnsTool-call traces

Retrieved Knowledge

Semantic vectors — metric and KPI definitions, table relationships, and past queries — retrieved to query the right way.

Vector search hitsKeyword matchesDoc snippets

Environment Results

Query results, row counts, and any SQL or permission errors — parsed to chart, explain, or safely retry.

API responsesFunction returnsErrors & status

Capabilities

Self-Service Analytics, Done Safely

Natural-Language Queries

Ask a question in plain language; the agent writes and runs the query for you.

Auto Charts

Picks the right visualization for each answer and renders it instantly.

Written Narratives

Explains what the data means in clear prose, not just numbers on a chart.

Semantic-Layer Aware

Uses your governed metric and KPI definitions so numbers match the source of truth.

Scheduled Reports

Runs recurring analyses and delivers narrated reports to stakeholders automatically.

Governed & Cited

Shows the query it ran and the sources behind every number, for trust and audit.

How it works

From Question to Insight

01

Ask

A user asks a question in plain language from chat or a dashboard.

02

Plan

The agent resolves it against your semantic layer and governed metric definitions.

03

Query

It writes and runs safe, correct SQL against your warehouse or lake.

04

Visualize

It picks and renders the right chart for the result.

05

Explain

It writes a narrative and shows the query and sources behind the answer.

Powered by the Vervelo Agents Platform

Open Source. On Your Infrastructure.

The Analytics Agent is built on our open-source platform — deployable in your cloud, on-prem, or air-gapped — so your data and queries never leave your environment.

Explore the Platform →

Governed, Secure Analytics

The agent respects your access controls and metric governance, logs every query for audit, and handles data safely — aligned to standards like SOC 2 and GDPR — and because you self-host, your data never leaves your environment.

SOC 2 GDPR
Vervelo company logo

Vervelo is a digital-health software partner blending deep clinical insight with world-class engineering to build tailored, secure, interoperable healthcare platforms.

Benefits of custom software solutions
  • Software delivered ownership benefit

    You fully own IT consulting and software delivered

  • Highly personalized solution benefit

    You get a highly personalized solution

  • Integration capability benefit

    Customize and integrate seamlessly

  • Scalability benefit

    On-demand scalability is always possible