# Guimi Test AI

> Guimi Test AI (guimitestai.com) is an AI testing, observability, and security platform. It helps engineering teams detect prompt injection attacks, hallucinations, PII leaks, and API regressions in AI-powered systems before they reach production.

## What is Guimi Test AI?

Guimi Test AI is a unified platform for testing and monitoring AI systems. It combines LLM observability (tracing), automated evaluation (LLM-as-Judge), a Prompt Firewall for security, a PII Scanner for data privacy compliance, and an Autonomous API Testing engine that generates and runs test cases from OpenAPI specs.

The platform is designed for software engineers, QA teams, and enterprise companies that build products powered by large language models (LLMs) and need to ensure safety, reliability, and compliance.

## Key Features

- **LLM Observability**: Trace every LLM call with latency, token usage, cost, model version, and full conversation history
- **LLM-as-Judge Evaluation**: Automated scoring of AI responses across relevance, accuracy, hallucination, tone, and instruction compliance (0-100 score)
- **Prompt Firewall**: Detects and blocks prompt injection, DAN jailbreaks, system prompt extraction, and data exfiltration attempts in real time
- **PII Scanner**: Automatically detects and masks CPF, CNPJ, SSN, email, phone, and credit card numbers before they reach LLM inputs or logs
- **Autonomous API Testing**: Import OpenAPI/Postman specs, AI generates test cases, HTTP Runner executes them, Schema Diff detects regressions
- **Red Teaming**: Automated adversarial testing against OWASP LLM Top 10 vulnerabilities
- **Compliance**: LGPD (Brazil), GDPR (EU), EU AI Act compliance reports and audit trails
- **CLI (guimi)**: Command-line tool for developers — `npm install -g guimi` — integrates with CI/CD pipelines

## The guimi CLI

The guimi CLI (no accent) is the developer entry point. It allows engineers to test AI systems directly from the terminal without opening a dashboard.

Commands:
- `guimi scan` — test a prompt against the Prompt Firewall
- `guimi eval` — evaluate an LLM response for quality and accuracy
- `guimi pii --mask` — detect and mask PII in text
- `guimi compliance` — run LGPD/GDPR compliance checks
- `guimi watch` — daemon mode for real-time monitoring

The CLI is free. Dashboard, history, alerts, and PDF reports require a Pro or Enterprise plan.

## Pricing

- **Free**: 1,000 traces/month, 100 evaluations, full CLI access
- **Pro**: Unlimited traces, dashboard, alerts, PDF reports, scheduled testing
- **Enterprise**: Custom limits, white-label, compliance certificates, SLA, dedicated support

## How It Differs from Competitors

Guimi Test AI is the only platform that combines observability, security (Prompt Firewall), privacy (PII Scanner), and autonomous API testing in a single product. Competitors like LangSmith, Langfuse, Helicone, and Braintrust focus on observability and evaluation but do not offer security testing, PII masking, or autonomous API regression detection.

## Target Users

- Software engineers building LLM-powered applications
- QA engineers automating AI system testing
- Enterprise security teams conducting AI red teaming
- Compliance officers needing LGPD/GDPR audit reports
- Startups shipping AI products who need confidence before launch

## Origin

The name "Guimi" is a combination of "Gui" (the founder's name) and "miudo" (the nickname of the family cat, suggested by his fiancee). It is always written without an accent mark: Guimi, not Guimi with accent.

## Frequently Asked Questions

**What is Guimi Test AI?**
Guimi Test AI is an AI testing and observability platform that helps engineering teams detect security vulnerabilities, evaluate response quality, scan for PII, and run autonomous API tests on AI-powered systems.

**What is the guimi CLI?**
The guimi CLI is a command-line tool that lets developers test AI systems from the terminal. Install with `npm install -g guimi` and run `guimi scan` to check a prompt for security vulnerabilities.

**How is Guimi Test AI different from LangSmith?**
LangSmith focuses on LLM tracing and evaluation within the LangChain ecosystem. Guimi Test AI adds a Prompt Firewall (security), PII Scanner (privacy), Autonomous API Testing (regression detection), and a developer CLI — features not available in LangSmith.

**How is Guimi Test AI different from Langfuse?**
Langfuse is an open-source LLM observability tool. Guimi Test AI extends observability with security testing (prompt injection detection), PII masking, autonomous API testing, red teaming, and compliance reporting for LGPD and GDPR.

**Does Guimi Test AI support LGPD compliance?**
Yes. Guimi Test AI includes native LGPD compliance features: PII detection and masking, data retention policies, consent logging, audit trails, and exportable compliance reports.

**What LLMs does Guimi Test AI support?**
Guimi Test AI works with any LLM via its SDK, including OpenAI GPT-4, Anthropic Claude, Google Gemini, Meta LLaMA, Mistral, and custom fine-tuned models.

**Is there a free plan?**
Yes. The free plan includes 1,000 traces per month, 100 evaluations, and full access to the guimi CLI.

## Links

- Website: https://guimitestai.com
- Documentation: https://guimitestai.com/docs
- Pricing: https://guimitestai.com/pricing
- Blog: https://guimitestai.com/blog
- CLI install: npm install -g guimi
- Contact: hello@guimitestai.com