⚡ Drop-In OpenAI-Compatible SaaS Gateway

Deterministic Edge Agent Guardrails at 0ms Token Overhead.

Connect your existing OpenAI, Groq, or local vLLM keys. SteerGuard applies zero-latency vector steering and AST self-healing at the edge. Replace expensive 1,000-token system prompts with Contrastive Activation Addition (CAA) steering vectors.

1,500 System Tokens

420ms

SteerGuard Edge

38ms

📚 Based on CAA & Representation Engineering Research (Rimsky et al., 2023)

Launch Live Playground

104,200

Requests Processed

99.96%

AST Verification Rate

112.4M

Prompt Tokens Saved

31.5 ms

Avg Edge Latency

🗺️ Developer Onboarding Architecture

4 Frictionless Onboarding Pathways (0s to 60s)

PATHWAY 1 • 0 SECONDS

Zero-Install WebGPU

In-browser forward-hook simulation powered by LiteRT.js & WebAssembly SIMD kernels.

Test Web Simulator

PATHWAY 2 • 0 SECONDS

Serverless Edge Proxy

Drop-in OpenAI-compatible gateway with lookahead buffering & AST self-healing.

https://api.paulcreates.online/v1

PATHWAY 3 • 60 SECONDS

1-Click Google Colab

Extract real CAA steering tensors (h_L^+ - h_L^-) on a free cloud T4 GPU and deploy to edge.

PATHWAY 4 • 30 SECONDS

1-Line CLI Wizard

Detects GPU/MPS hardware, trains local vectors, and audits prompt waste.

steerguard init

🔬 Real GPU Representation Extraction

Train Your Custom Steering Vector (Free Colab T4 GPU)

🚀 Open Training Notebook

Extract mathematical representation differences (h^+ - h^-) on an un-gated SLM (Qwen 2.5 1.5B or Gemma 2 2B) and deploy the compiled .bin directly back to this edge gateway via 1-click webhook.

STEP 1: GENERATE SESSION TOKEN

STEP 2: EDGE WEBHOOK LISTENER

⏳ Ready to receive trained vector tensors

Live Interactive Simulator

Representation Steering Playground

⚡ Powered by Cloudflare Workers AI + SteerGuard Edge AST Guardrail

✎ PROMPT:

SSE Stream

❌ Vanilla Baseline (α = 0.0) AST UNSTEERED

• Burning 1,000+ context tokens for system prompt instructions

✅ SteerGuard (Layer 18 Hook) AST VERIFIED

• 0 prompt tokens burned | 42ms TTFT Saved: 1,024 tokens

Cost & Latency Reduction

System Prompt Token ROI Calculator

Monthly API Requests 500,000

System Prompt Tokens Saved / Req 1,200 tokens

Model Token Pricing (per 1M tokens) $2.50

Estimated Monthly Savings

$1,500 / mo

Eliminates 600,000,000 prompt tokens per month with 0ms prefill lag.

Commercial Self-Serve

Transparent Tier Pricing

Community

$0 / month

✓ Local Python runtime
✓ 36 compiled steering tensors
✓ Full CLI access
✓ Public community support

Dev Pro

$49 / month

✓ 100k edge proxy requests (BYOK)
✓ 9 curated domain bundles
✓ AST auto-repair engine
✓ Lookahead stream buffer

Team / Scale

$149 / month

✓ 500k monthly requests
✓ 5 team seats included
✓ Automated DPO dataset export
✓ Webhook violation alerts

Enterprise

$999+ / month

✓ Custom vector compilation
✓ On-premise air-gapped WASM core
✓ Dedicated SLA included
✓ HIPAA BAA & SOC2 compliance