Vericto S.A.S.
Vericto provides a deterministic database-security layer for AI agents and other software that generates SQL. Its AST-based engine evaluates statements before execution, blocking destructive queries, masking sensitive columns, enforcing agent access policies, and creating an auditable explanation and fix.
At a Glance
- Engineering and platform teams deploying AI agents
- Security and compliance teams
- Database administrators and data teams
- SaaS companies with production PostgreSQL or MySQL
- +2 more
AI Tools by Vericto S.A.S.
(1)
Vericto Proxy
SQL Guardrail Proxy for Databases
Discussions
No discussions yet
Be the first to start a discussion about Vericto S.A.S.
Latest News
Published the practical guide ‘How to implement database guardrails for AI agents’.
Published a deep dive on building a PostgreSQL proxy with libpg_query and the wire protocol.
Published the official benchmark of proxy latency and resource overhead.
Published a text-to-SQL guardrails guide for preventing destructive or over-broad LLM queries.
Products & Services
Self-hosted PostgreSQL/MySQL wire-protocol proxy that parses queries into ASTs and blocks or masks them before they reach the database, without application code changes.
API for evaluating queries from application code before execution across six supported SQL dialects; the application honors the verdict.
Command-line SQL validator for pre-commit and CI/CD, with SARIF/GitLab output, rule inspection, offline checks, and signed receipt verification.
Batch SQL-check API for CI/CD that returns a verdict and exit code without executing statements.
Market Position
Vericto positions itself as a deterministic AST-based database guardrail rather than a regex/keyword filter, a simple operation allowlist, a database-role permission system, or an ML-based detector. Its differentiators are enforcement inline on the database wire path, actionable rule/fix explanations, multi-channel checks for agents and CI, sensitive-column masking, and auditable deterministic decisions. The official site directly compares it with database roles and SQL allowlists.
Founding Story
Vericto says it was born from the question of why AI should be used to protect against AI when the problem has a mathematical solution. Its initial vision was to provide a deterministic AST security layer between LLMs and production databases—analogous to HTTPS as a standard security layer—so engineering teams could deploy database-connected agents with the confidence of human-reviewed code.
Business Model
Revenue Model
Subscription SaaS priced by protected databases, query volume, features, support, and enterprise deployment/SLA; self-hosted proxy and source-available components are used as part of the paid control-plane service.
Pricing Tiers
1 database, 500K queries/month, 42 standard rules, 7-day history, and 1,000 CLI/MCP checks/month.
3 databases, 5M queries/month, custom YAML rules, 30-day history, audit export, and 10,000 CLI/MCP checks/month.
10 databases, 50M queries/month, 20 custom rules, 90-day history, signed PDF report, near-real-time alerts, and unlimited CLI/MCP checks; extra databases are $25/month each.
Annual contract with unlimited queries, SSO/OIDC, VPC deployment, SLA/service credits, compliance support, contract-set retention, and 24/7 P1/P2 support.
Target Markets
- Engineering and platform teams deploying AI agents
- Security and compliance teams
- Database administrators and data teams
- SaaS companies with production PostgreSQL or MySQL
- Teams using MCP, LangChain, LlamaIndex, dbt, Airflow, GitHub Actions, or GitLab CI
- Enterprises requiring SSO, VPC deployment, SLAs, and audit evidence
- Protecting production databases accessed by autonomous or coding AI agents
- Text-to-SQL assistants and internal admin tools
- MCP database servers and LangChain/LlamaIndex SQL agents
- CI/CD validation of migrations and generated SQL
- Pre-commit SQL checks and GitHub/GitLab pipeline enforcement
- Data pipelines and orchestration with dbt and Airflow