Talk to Your Data. Stop Writing Queries, Start Asking Questions.

What if anyone in your organization could ask questions of company data—enrollment records, billing transactions, course schedules- just by asking a question in plain English? With Actian AI Analyst and the Actian Zen JDBC connector, that future is already running in production.

THE IDEA 

From a JDBC Connection to Conversational Intelligence

The Actian Zen database engine has long powered transactional workloads across education, finance, and enterprise-fast, embedded, and JDBC accessible. The missing piece was always a natural language layer that could reason over that data without requiring SQL expertise. 

AI Analyst fills that gap. Connect your Zen JDBC data source once, let the Steward Agent (the AI model builder) explore your schema and draft a semantic layer automatically, and your users can start asking questions immediately. No ETL, no warehouse migration, no data team backlog. 

Here are three steps to get started: 

GETTING CONNECTED 

Step 1. Wire Up Your Zen JDBC Connection

Everything starts in the Connections panel. Actian AI Analyst accepts a standard Zen JDBC connection string, the same format your existing apps already use. No custom configuration.  

# JDBC Connection String (without jdbc:pervasive:// prefix)

host    : ec2-23-432-123-238.us-east-2.compute.amazonaws.com

port    : 1583

database: Demodata

uid     : Master

pwd     : ••••••••

# After saving: click 'Test Connection' to verify

Once connected, AI Analyst can read your schema, discover all tables, and automatically draft semantic models for review and refinement. 

 MODEL CREATION 

Step 2. Let Steward Build Your Semantic Layer

Steward automates the creation of semantic models within Actian AI Analyst. When you click + New Model, you choose your creation method. Steward (recommended) analyzes your schema, reads business context, and drafts an initial model—dimensions, measures, relationships, and agent guidance—entirely on its own. 

In a real session against the Zen Demodata database, Steward was given a single instruction: “all of those tables.” It then executed 11 parallel SQL queries, explored each table’s structure, and generated an initial semantic layer—all within a single Steward conversation, with no human SQL written. 

STEWARD AI—ACTUAL SESSION OUTPUT 

“Great choice! I’ll create models for all 11 tables. Let me start by fetching the guide and exploring each table’s structure.”

— Steward, after the user typed: “all of those tables”

THE SEMANTIC LAYER 

What Steward Built—Automatically

From a single Zen JDBC connection to a university demo database, Steward produced 11 fully modeled tables. Each model includes dimensions, measures, cross-model relationships, and AI agent guidance.  

students → person, enrolls → classes → faculty → dept—makes cross-model analysis seamless for your AI analysts without any manual configuration. 

THE AI ANALYST 

Step 3. Configure and Deploy Your Analyst

With models in place, you configure your AI Analyst to align with the intended end user, write optional instructions for domain context, and use the built-in test panel to validate queries before deploying. 

• 10 models connected. Selectively assign which tables your analyst can access. 
• 15 starter suggestions. Auto-generated example questions help new users know what to ask. 
• One-click deployment. Share via a link, embed in your portal, or expose via API. 
• Usage monitoring. Track who’s asking what and where the semantic layer needs refinement.

CONVERSATIONAL AI IN ACTION 

Plain English. Real Results.

Here’s what happens when a user asks a question; no SQL or BI tool knowledge is needed. The AI Analyst queries the semantic layer, identifies the correct join path, and returns structured results.  

Example 1. Financial Query: Students with Outstanding Balances

User prompt: “List the first 5 students who owe more than $2,000” 

Behind the scenes, the analyst identified billing_transactions as the source of truth, joined it with students and person tables, and computed outstanding balances—from five reasoning steps, no human SQL involved. 

Example 2. Cross-Model Enrollment Query: Calculus II Students

User prompt: “List the students who enrolled in Calculus II” 

 CROSS-MODEL JOIN—HOW THE ANALYST THINKS 

“I need to join course, classes, and enrolls to get students enrolled in Calculus II, then join person for their names. Let me check the relationships to use the correct join keys.”

Join path: enrolls → classes via class_id → id · enrolls → students via student_id → id · students → person via id → id · Executed in 5.8s 

WHY IT MATTERS 

The Semantic Layer is the Key

Traditional BI tools require someone to build reports before anyone can see data. Actian AI analysts flip that model. The semantic layer is built once by Actian AI Analyst, and then anyone can explore it freely. 

• Semantic SQL, not raw SQL. The AI writes against your semantic models with business logic baked in. Fewer errors, consistent results. 
• Model-level access control. Choose exactly which models each analyst can see. Security by design. 
• Cross-table reasoning. Relationships defined once mean the AI knows which join path to use. 
• Agent guidance per model. Add plain English notes to any model and the AI Analyst applies that context to every question. 

GETTING STARTED 

Connect Zen to AI Analyst 

If you’re running Actian Zen today, you’re one JDBC connection string away from conversational analytics. The path is straightforward: 

For a typical Zen database with 10 to 15 tables, Steward can have a fully working analyst ready in less than 30 minutes. Once Steward drafts the semantic layer, you can refine it, add custom measures, adjust descriptions, and set agent guidance. 

Actian AI Analyst is available through its web application and includes a 14-day free trial. Connect your first Zen data source at Connections → + New Connection → Zen JDBC.  

CONCLUSION 

Beyond the Query: What the Semantic Layer Actually Does

What these benefits illustrate is something deeper than a chatbot on top of a database. The real innovation in Actian AI Analyst is the semantic layer, a structured, AI-readable description of your data that sits between raw SQL and natural language. 

Once that layer exists, every user in your organization becomes capable of querying your Zen database directly. A university administrator can ask which students owe the most tuition. A registrar can find everyone enrolled in a specific course. A department head can see which faculty are teaching class overloads, all without opening a BI tool, filing a ticket, or writing a line of SQL. 

For organizations already running Actian Zen, the barrier to entry is intentionally small: one JDBC connection string, one Steward session, and you have a working AI analyst. There’s no data warehouse to provision, no ETL pipeline to maintain, no BI layer to license separately. 

Your data has always had the answers. Now anyone can ask them.

About Author

About Johnson Varughese

Johnson Varughese manages Support Engineering at Actian, assisting developers leveraging Actian data interfaces (ODBC, JDBC, ADO.NET, and more). As a Google Certified Data Practitioner and Generative AI Leader, he provides technical guidance and troubleshooting expertise to ensure robust application performance across different programming environments. Johnson's wealth of knowledge in data access interfaces has streamlined numerous development projects. His Actian blog entries detail best practices for integrating Actian’s database engines and interfaces. Explore his articles to optimize your database-driven applications.