Visual Expert AI

AI-assisted code analysis and documentation for Oracle PL/SQL, SQL Server T-SQL, and PowerBuilder — built on top of Visual Expert's static analysis engine.

Discover VE AI features ↓ Security & Architecture ↓ Using VE AI ↗ FAQ & Security ↗

Features

Visual Expert AI features

Five AI features, each designed for a specific code analysis or improvement task. Each can be used independently or in combination.

Explain the Business

Plain-language explanation of what a procedure or method does — designed for business analysts and project leads. Focuses on functional role, not implementation details. Learn more.

Non-technical

Explain the Logic

Technical summary of the code's inner logic, for developers who need to understand unfamiliar code quickly. Covers control flow and data handling without low-level noise. Learn more

Developer

Optimize Code

Performance, readability, and maintainability suggestions. Proposes more efficient logic, modern practices, and concise rewrites — reducing technical debt incrementally. Learn more

Refactoring

Generate Comments

Inserts context-aware inline comments automatically, improving documentation quality and long-term maintainability — without manual effort. Learn more

Documentation

Fix Inspection Issues

Proposes concrete fixes for problems flagged during static code analysis — paired with VE's inspection rules for immediate, actionable remediation. Learn more

Code review

Important: As with any AI service, results should be reviewed by a human before use. VE AI makes suggestions — you decide. Repeating the same question may produce different (but equally valid) phrasings. Response time may reach a few minutes for complex procedures.

Security & data handling

How your data is protected

Read-only

VE never modifies your source code, database schemas, or any element in your environment.

Minimal excerpts

Only the specific code fragment needed for your question is sent to the AI service — not full repositories.

Encrypted in transit & at rest

TLS in transit, AES at rest. Tenant isolation enforced in VE Cloud.

No model training

Your code is never used to train or fine-tune AI models. Data deletion on request within 30 days.

Admin controls

Disable AI per project by setting credits to zero. Block outbound traffic to AI endpoints if required.

Human in the loop

AI output is always a suggestion. The developer reviews and decides whether to apply it.

Detailed articles on how VE AI handles your data, manages access, and integrates with your infrastructure.

Security, confidentiality, and technical matters

U1 What does VE analyze and does it modify our systems?: VE analyzes only selected code and database object definitions and does not modify systems — it operates read-only.
U2 What exactly is sent to the AI when I run a feature?: Only your request and the strictly necessary code excerpt — never a full codebase or production data.
U3 What is Project Context and where is it kept?: Project Context is a per-tenant knowledge layer kept in your VE Cloud tenant and may be retained by OpenAI (via API) to improve future answers.
U4 Can we request deletion and how long does OpenAI keep data?: You can request deletion at any time. OpenAI deletion occurs within 30 days after instruction or at end of service.
U5 Are inputs or outputs used to train AI models?: No. Inputs and outputs are not used to train or fine-tune any AI model.
U6 How is the data protected end-to-end?: TLS in transit and AES encryption at rest, with tenant and project isolation in VE Cloud.
U7 Which network prerequisites must we enforce?: Allow outbound HTTPS (port 443) to ai.visual-expert.com with DNS resolution enabled.
U8 Can we operate VE without AI or temporarily block AI features?: Yes. VE works fully without AI. Admins can set AI credits to zero or block outbound traffic to AI endpoints.
U9 Can anyone at Novalys access our code?: No. Access by Novalys support staff requires explicit written authorization from your organization and is strictly limited to the data needed to resolve your support case.
U10 What are our responsibilities in the shared model?: Limit analysis scope, avoid secrets in excerpts, control who runs AI features, and request deletion of project context when needed.