Design Partner Pilot Package

SKU: QuantZK Control: AI Spend Firewall

Put hard, cross-provider financial controls on AI agents before they spend money, and automatically catch billing errors afterward.

Cloud IAM controls what machines can access. QuantZK controls what machines can spend.

Offer

QuantZK Control: Design Partner Pilot (30 to 60 days, paid $5k to $15k)

Not sold as Charge Authority, QZR, VDI, or Groth16. Those stay behind the product.

Includes:

• credential-isolated gateway, customer VPC preferred
• workload identities plus per-agent and cost-center budgets
• per-request hard caps plus vendor and model allowlists
• real-time ALLOW/DENY plus atomic reservations
• cross-provider spend aggregation, OpenAI and Anthropic minimum
• reconciliation (AUTHORIZED / EXPECTED / OBSERVED / BILLED)
• Spend Operations dashboard (/spend) plus evidence export
• weekly Protected Value and ROI report

Buyer and ICP

Primary: Head of AI Platform / AI Infrastructure
Secondary: FinOps / Cloud Cost, with Security as champion assist

ICP: multi-provider, production agents, meaningful usage spend, central platform team, typically $100k to $1M+ per month automated LLM or API spend. Not a $3k per month OpenAI-only startup.

Scope

1 business unit
1 to 2 providers
5 to 25 workloads or agents
one real monthly budget
one protected gateway deployment
reconciliation dashboard
weekly ROI report

Success criteria

>$X machine spend controlled
100% protected requests gated
zero budget oversubscription
denied requests → zero upstream calls
>95% spend reconciled
billing anomalies surfaced
integration <1 to 2 engineering days

Recommended deployment

Customer VPC

  agent / LiteLLM / OpenAI SDK
           │  QuantZK workload token
           ▼
  QuantZK Gateway (self-hosted)     ← provider keys stay here
           ├── OpenAI
           └── Anthropic
           │
           ▼ minimum control / reconcile state
  QuantZK Control → /spend + evidence

Integration surfaces

Integration	How
OpenAI	base_url=…/v1 + QuantZK workload token
Anthropic	POST …/v1/messages + QuantZK bearer (gateway injects x-api-key)
LiteLLM	OpenAI-compatible api_base → QuantZK /v1

Killer demo

cd protocol && npm run demo:ai-spend-firewall

Concurrent shared budget, settle and release, cross-provider policy, unauthorized DENY, then OVER_BILLED in reconciliation.

Pricing stance

Do not price on cryptographic proofs. Do not lock annual pricing before pilots. Annual later can key off controlled spend, workloads, or enterprise tier after observing real usage.

Explicit non-goals for pilot

• Replacing provider billing portals
• Becoming a PSP or Metronome
• Claiming QZR proves no bypass, Claim 4 stays intact
• Shipping CA v2 or delegated-budget cryptography before customers demand portable proofs

Commercial CTA

• Pilot page: quantzk.com/design-partner
• Book time: cal.com/omar-quantzk/30min
• Contact: contact@quantzk.io