SS/AP
SSAP
Decision system
Decision pathsGovernanceLatency controlQuality monitoringCost stability

A decision system for modern AI.

Decides when AI should act, when to try cheaply, and when not to act at all.

Built for governance: no raw prompts or completions stored by default. Or email a pilot request

Ways to use SSAP

InferenceGate and SupportGate are concrete decision surfaces — not separate products. Same SSAP underneath, different inputs and constraints.

Use case

InferenceGate

Decision control for AI inference
Production
  • • Controls when full inference is justified
  • • Prevents unnecessary or risky AI execution
  • • Stabilizes cost, latency, and behavior
Learn more →
Use case

SupportGate

Decision layer for AI-powered support
Support
  • • Decides which tickets need full AI
  • • Reduces AI usage without changing UX
  • • Designed for production support flows
Learn more →

How SSAP works (high-level)

SSAP is a decision architecture: it makes an explicit call about execution — when AI should act, when a cheap attempt is enough, and when no action is the correct outcome.

Where SSAP fits
SSAP sits above execution. InferenceGate / SupportGate are two common surfaces, but the decision system is the same: policy → decision → outcome.
Your App / ClientsSame API contractSSAPDecide → Try cheap → Or don’t actNO_ACTION / LIGHT / FULLExecutionLLMs / tools / flowsTelemetry & governance (no raw prompts)Optional: Shadow QA (quality evidence)
NO_ACTION
Don’t run AI. Return deterministic output.
LIGHT
Try cheaply first (bounded risk).
FULL
Escalate only when justified.
Governance
Structured telemetry you can audit.
Web vs PDF: This page explains what SSAP decides. The PDF explains how it decides (SS → AP), decision paths, telemetry, and deployment.
Decision 1
Should AI act at all?
Policy allows NO_ACTION as a first-class outcome.
Decision 2
Can we try cheaply first?
LIGHT attempts are bounded and auditable.
Decision 3
When is FULL justified?
Escalate only when policy and confidence demand it.
The PDF covers SS → AP, decision paths, telemetry, and deployment.

What happens after you apply

A short fit check, a scoped pilot, then a clear go / no-go decision backed by telemetry. Pricing is shared after we confirm pilot fit and success metrics.

Step 1
15-minute fit check
Confirm your surface (InferenceGate / SupportGate), traffic, constraints, and success criteria.
Step 2
Pilot setup
Drop-in decision layer with deterministic decision paths and telemetry. Minimal disruption.
Step 3
KPI report + decision
You get a report (outcomes, latency, quality signals, governance). Then you decide if rollout makes sense.
Make AI execution an explicit decision
Govern when AI acts, when it tries cheaply, and when it shouldn’t act at all.

Send a message

Send a direct note. Please include your email, company, and your approximate monthly AI/LLM spend so we can respond with the right pilot path.

Contact details
Or email directly:
marko@ssap.io
What we reply with
  • • Suggested pilot surface (InferenceGate / SupportGate)
  • • Governance + compliance considerations
  • • Latency + quality impact expectations
  • • Integration notes (drop-in)
We’ll use your email only to reply.

FAQ

What does SSAP decide?

It decides whether AI should act, whether to try a cheap path first, or whether no action is the correct outcome.

Are InferenceGate and SupportGate separate products?

No — they’re two common decision surfaces. Same SSAP underneath, different inputs and constraints.

Is any prompt content stored?

No raw prompts or completions by default. Telemetry is structured and audit-friendly.

What is NO_ACTION?

A first-class outcome: the decision system returns a deterministic response shape without invoking a model.

Why is the PDF gated by email?

The PDF is opt-in depth, and we use the email only to deliver the document and reply if you request a pilot.

What happens after the pilot?

You keep the KPI report and decide go / no-go.