Skip to main content
PromptWall offers three modes depending on who owns the LLM tokens:

Verify Only

You already have an answer. We validate it.

Webhook BYOK

You bring your LLM API key. We run the full pipeline.

Webhook Managed

We provide the LLM. You just send prompts.

​
1. Verify Only

Flow: 
Your App → /v1/verify {prompt, answer, tool_result} → PromptWall
    ← {governance, confidence, evidence_consistent}
No LLM call is made by PromptWall. We validate the answer against the tool_result (or against detected internal knowledge). Fastest and cheapest. When to use:
  • You already call the LLM yourself and want an independent sanity check
  • You’re adding governance to an existing pipeline with minimal surface-area change
  • You want lowest latency (~150ms p95)
Pricing structure: flat monthly fee + small overage per request.

​
2. Webhook BYOK (Bring Your Own Key)

Flow: 
Your App → /v1/chat {prompt}
           → PromptWall: scanner → policy → grounding → your tool webhook
           → PromptWall → LLM (using YOUR OpenAI/Anthropic key) → judge → enforce
           ← {answer, governance, ...}
You store your LLM API key encrypted in PromptWall (Fernet + KMS). We use it to call the LLM on your behalf. You pay the LLM provider directly. When to use:
  • You already have provider contracts (OpenAI enterprise, Azure, etc.)
  • You want full pipeline: scan → LLM → verify → audit
  • You don’t want PromptWall markup on tokens
Pricing structure: flat monthly + per-request overage. LLM tokens billed by OpenAI/Anthropic directly to you.

​
3. Webhook Managed

Flow: 
Your App → /v1/chat {prompt}
           → PromptWall runs everything including LLM call with OUR key
           ← {answer, governance, ...}
PromptWall owns all LLM costs. We charge you per request + markup on tokens. When to use:
  • You want the simplest integration (no LLM provider setup)
  • Variable load where fixed commitment doesn’t make sense
  • Team doesn’t want to manage multiple vendor relationships
Pricing structure: per-request + token markup (1.15x - 1.50x based on tier).

​
Comparison

VerifyBYOKManaged
LLM tokensn/aCustomer pays providerPromptWall
Integration complexityLowMediumLow
Latency~150ms~800ms~800ms
Our cost per request$0.0001$0.001$0.001 + tokens
Best forExisting pipelinesEnterprise with provider contractsStartups, variable load

​
Switching modes

A single tenant can use multiple modes via different API keys:
  • Create a new “app” in the dashboard
  • Choose the mode per-app
  • Point your code at the appropriate API key
Usage, billing, and limits are tracked separately per app.