Platform · API Gateway

Server-side AI calls. Filtered just like the browser.

The Chrome extension protects what your team types. But what about internal tools that call OpenAI or Anthropic directly from your stack? The Ventrin API Gateway proxies server-side requests through the same policy engine — before the prompt reaches the model.

Join Early Access REST API reference

<10ms

Added latency overhead on proxied requests — measured at the 99th percentile under normal load

Line of code to integrate — change your OpenAI client's base URL to api.ventrin.com/v1. No other changes required.

Unified

Server-side API calls appear in the same audit log as browser interceptions — one compliance trail for your entire AI surface

Integration

One URL change. Full coverage.

Drop-in proxy — one line to integrate.

Change your OpenAI or Anthropic SDK's base_url parameter to https://api.ventrin.com/v1. All requests are transparently proxied — the response format is identical to the provider's native API. No other SDK changes, no new dependencies, no refactoring. Works with every OpenAI-compatible client library.

Same policy engine as the extension.

Policies you've configured in the admin dashboard apply to API calls automatically. There's no separate ruleset to maintain for server-side traffic. If a policy says client names should be redacted before reaching external AI tools, that applies whether the request comes from a browser or a backend service. One rule, everywhere.

Unified audit trail across channels.

Server-side API calls generate event log entries in exactly the same format as browser interceptions. Your compliance dashboard shows the full picture — ChatGPT sessions from the marketing team, Claude API calls from the legal tech platform, Gemini queries from a document automation tool — all in one place, all filterable and exportable.

The coverage gap

The extension covers browsers. This covers everything else.

Law firms and regulated businesses increasingly build internal tools on top of foundation models — document drafting assistants, matter summarisers, due diligence accelerators. These tools make direct API calls to OpenAI or Anthropic from your stack. The Chrome extension doesn't intercept them. The API Gateway does.

Works with LangChain, LlamaIndex, and any OpenAI-compatible client library
Authentication via API key — scoped to your Ventrin workspace
Streaming responses fully supported — no buffering penalty
Fallback mode available — if gateway is unreachable, requests can be configured to pass-through or fail-closed

Gateway integration · Python

client = OpenAI(

base_url= api.ventrin.com/v1

)

Policy applied WORKSPACE

Audit log UNIFIED

Added latency <10ms

One-line integration — before & after

Before Python · direct OpenAI call

import openai

client = openai.OpenAI(
  api_key="sk-…",
  # base_url not set → hits OpenAI directly
)

response = client.chat.completions.create(
  model="gpt-4o",
  messages=[{
    "role": "user",
    "content": prompt   # ⚠ unscanned
  }]
)

After Python · through Ventrin gateway

import openai

client = openai.OpenAI(
  api_key="sk-…",
  base_url="https://gateway.ventrin.com/v1",
  # ↑ that's the only change
)

response = client.chat.completions.create(
  model="gpt-4o",
  messages=[{
    "role": "user",
    "content": prompt   # ✓ scanned + logged
  }]
)

Before Node.js · direct call

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.OPENAI_KEY,
  // baseURL not set
});

const res = await client.chat.completions
  .create({ model: 'gpt-4o',
    messages: [{ role: 'user',
      content: prompt }] });  // ⚠ unscanned

After Node.js · through gateway

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.OPENAI_KEY,
  baseURL: 'https://gateway.ventrin.com/v1',
  // ↑ that's the only change
});

const res = await client.chat.completions
  .create({ model: 'gpt-4o',
    messages: [{ role: 'user',
      content: prompt }] });  // ✓ scanned

Before curl · raw API

curl https://api.openai.com/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role":"user","content":"..."}
    ]
  }'

After curl · through gateway

curl https://gateway.ventrin.com/v1/chat/completions \
  -H "Authorization: Bearer $OPENAI_KEY" \
  -H "X-Ventrin-Key: $VENTRIN_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [
      {"role":"user","content":"..."}
    ]
  }'
# same response shape · +<10ms overhead

The gateway is OpenAI-API compatible — any SDK that supports a custom baseURL works without further changes. Anthropic, Gemini, and Mistral endpoints also supported. See API docs for full reference.

Close the server-side gap.

One base URL change. Full policy coverage for your entire AI surface.

Join Early Access

View API reference →

OpenAI + Anthropic compatible · <10ms overhead · Unified audit log.