[AI gateway for developers]

The gateway for every AI call your app makes.

Inspect, debug, protect, and control every prompt before it reaches an LLM. Drop-in SDK. Open-source. EU-hosted.

Get Started FreeView Docs
// One line to observe, debug, and protect every AI call
const grepture = new Grepture({
apiKey: "gpt_abc123",
proxyUrl: "https://proxy.grepture.com",
});

const client = new OpenAI(
grepture.clientOptions({
apiKey: "sk-openai-key",
baseURL: "https://api.openai.com/v1",
})
);
0M
Requests observed / wk
0M
Tokens tracked / wk
0K
PII fields redacted / wk
0.00%
Uptime (30d)

How it works

01

Connect

Point your SDK at Grepture. One line of config, zero code changes.

02

Observe

Inspect prompts, trace conversations, see token usage and cost per request.

03

Protect

Sensitive data is automatically detected and redacted before it reaches external providers.

Works with

OpenAIAnthropicGoogle AIAzure OpenAICohereMistralAWS BedrockHuggingFaceGroqReplicateOpenAIAnthropicGoogle AIAzure OpenAICohereMistralAWS BedrockHuggingFaceGroqReplicate

Capabilities

[PROMPT INSPECTOR]

See exactly what your LLM sees

Structured conversation view with before/after diff for every request. Replay any prompt to debug issues. Full visibility into what your AI actually receives.

[COST TRACKING]

Know what your AI costs

Token breakdown per request with per-model cost estimation. Spot expensive prompts, compare models, and track spending across your entire AI stack.

[CONVERSATION TRACING]

Follow the full conversation

Trace IDs link every request in a multi-turn conversation. Group agent loops, debug chain-of-thought, and see exactly how conversations flow across your system.

[DATA PROTECTION]

PII, secrets, and prompt injection — caught automatically

Names, emails, API keys, and adversarial inputs are detected and redacted before they reach any external provider. Configurable rules, reversible redaction, zero-data mode.

Your command center

Real-time visibility into every request.

app.grepture.com/traffic-log
Grepture dashboard showing prompt inspector, token costs, conversation tracing, and security detections

Inspect prompts, track token costs, trace conversations, and see which security rules fired — all in one dashboard.

Zero retention

[ZERO DATA MODE]

Full protection. Zero stored data.

Enable zero-data mode and Grepture processes every request — detecting PII, redacting secrets, blocking threats — without ever writing your content to disk. Headers, bodies, and URLs never touch our database. Only operational metadata is logged.

  • Rules still fire — PII detection, redaction, blocking, and tokenization all work in-flight
  • Only method, status code, latency, and rule hits are stored
  • One toggle in your dashboard. Instant. No migration needed.

Built for trust

[EU-HOSTED]

Your data never leaves Europe.

All Grepture infrastructure runs in the EU. Every subprocessor — database, cache, analytics, payments — is hosted in Germany or Ireland. GDPR-ready by default.

  • +All infrastructure hosted in Frankfurt & Nuremberg
  • +Every subprocessor EU-based — no US data transfers
  • +GDPR and EU AI Act ready out of the box
  • +Zero-data mode: nothing written to disk
[OPEN SOURCE]

Don’t trust a black box with your data.

The Grepture gateway is fully open source. Every detection rule, every redaction action, every byte of data handling is auditable. Self-host for full infrastructure control.

  • +Full gateway source code on GitHub
  • +Every detection rule readable and auditable
  • +No black-box processing — deterministic and transparent
  • +Self-host option for teams that need full control
View on GitHub →

Use cases

[]AI-Powered Apps Chatbots, summarizers, AI features[]AI Agents Autonomous agents, tool calls, MCP[]RAG Pipelines Knowledge bases, document chunks[]Prompt Debugging Inspect, replay, diff every prompt[]Cost Tracking Token costs, per-model breakdowns[]Multi-Model Gateway One dashboard, every provider[]Any API Call Webhooks, payments, any HTTP[]Shadow AI Protection Browser extension for ChatGPT

Start observing your AI traffic in 5 minutes

Drop-in SDK. See your first request in under a minute.

Free for up to 1,000 requests/month · No credit card required

Get Started Free