NEW: CHAT WITH ANY LLM PROVIDER

The best way to secure your
LLM apps, verify outputs,
and deploy enterprise AI.

Proxy any application, to any provider. Verify across multiple agents. Get reliability in your stack, in one click.

Free for internal testing. No card required.

~ curl -X POST https://proxy.groundkeeps.in/v1/chat/verify \
  -H "Authorization: Bearer gk_live_jwt..." \
  -d '{
    "messages": [...],
    "sources": [...]
  }'

// Response
{
  "status": "success",
  "hallucination_score": 0.0,
  "escalation_action": "pass",
  "audit_hash": "e3b0c442...b855"
}

The Middleware Architecture

A seamless proxy integrating Hallucination Detection and Agent Governance.

Your App
Ground Keeps Trust Proxy
Grounding Engine
RAG Reliability
Guard Engine
Agent Governance
LLM Providers (OpenAI, Anthropic, Ollama...)

Three Core Engines

Grounding Engine

RAG reliability solved. Uses a reasoning LLM to verify factual claims against sources, bypassing the 100% false-positive rate of embedding-based cosine similarity.

  • Claim Extraction & Verification
  • Source Ranking (Authority/Freshness)
  • Conflict Resolution

Guard Engine

Agent identity and governance via HMAC JWT tokens and declarative allow/deny glob policies.

  • Agent Identity Management
  • Declarative Glob Policies
  • < 0.01ms Verification Latency

Audit Store

Tamper-evident log ensuring regulatory compliance (EU AI Act, HIPAA). An append-only SQLite WAL log with a continuous SHA-256 Hash Chain.

  • SHA-256 Hash Chain Integrity
  • Filterable Queries
  • Pure JS SQLite (Zero native deps)

End-to-End Governance

Protecting your application before and after the LLM call.

Output Governance

Citation checking, regex-based PII scanning, and custom pattern blocking to catch obvious issues before grounding.

Document Governance

Pre-filters source documents by status, effective/expiry dates, supersession rules, and versioning.

Escalation Engine

Post-verification actions: Pass, Flag, Block, or Correct (re-prompt LLM to fix the response).

8 Supported Providers

OpenAI Anthropic Google Gemini Groq DeepSeek Together OpenRouter Ollama

Zero vendor lock-in. Switch via a unified API.

Simple Pricing

Start free. Scale when you need to.

Self-Hosted

Free

Open-source with rate limits. Run on your own infrastructure.

  • Full grounding & guard engine
  • Audit store with SHA-256 chains
  • React dashboard
  • 5,000 verifications / month
  • 2 agents
  • 15-day audit retention
  • Community support (GitHub)
Self-Host Now →
Popular

Self-Hosted Pro

$99/mo

International

or ₹8,499 India

Unlimited self-hosted for production use. Same infra, no caps.

  • Unlimited verifications
  • Unlimited agents
  • 1-year audit retention
  • Priority GitHub issues
  • License key (offline-capable)
  • Email support
Buy $99/mo → Buy ₹8,499/mo →

Enterprise

Custom

On-prem deployment, SSO, compliance, and dedicated support.

  • Everything in Self-Hosted Pro
  • On-prem / air-gapped deployment
  • SAML / SSO integration
  • SOC 2 evidence reports
  • Fine-tuned verification models
  • Dedicated support engineer
Contact Us →

Every LLM needs a guard.
GroundKeeps is yours.

Free to self-host. MIT licensed. No data leaves your infrastructure.

Start Free →