Your AI did something in production. Can you prove what?

The AI proposes. A deterministic core disposes. Every response carries trust boundaries, every action passes a confirmation gate, every output is traceable to its source. Raptor is the governance layer for AI systems that face auditors, not just users.

OMB M-25-21 · EU AI Act Art. 9–17 · SR 11-7 · HIPAA · IL4/IL5
Where Raptor sits

Three approaches. One actually governs.

Observability

Tells you what happened after.

Logs, traces, dashboards. Useful for debugging. Useless for the regulator asking what your AI was permitted to do.

Evals

Tests capability against known cases.

Measures whether the model can get it right. Doesn't enforce that it must, or prove that it did, in production.

Raptor

Governs what's permitted to happen.

Every response is validated against evidence requirements, marked with trust boundaries, and produced by the same deterministic process. Auditable by design.

How Raptor works

AI proposes. Deterministic core disposes.

The AI does what AI is good at — interpreting fuzzy input, generating candidates, inferring intent. The deterministic layer does what rules are good at — validating those candidates against your evidence requirements, applying your governance rules in the same order every time, marking trust boundaries based on actual provenance, and producing the output your users act on.

Proposal Layer

  • AI interprets fuzzy input
  • AI generates candidates
  • AI infers intent
  • Full upstream flexibility

Commit Layer

  • Rules validate evidence
  • Governance applies
  • Trust boundaries mark output
  • The output users act on

01 Every response is structured

Each segment carries a trust boundary — CONFIRMED EXECUTED RETRIEVED INFERRED UNCERTAIN. Visible inline, backed by response-level provenance.

02 Every action is governed

When the AI proposes an action, Raptor enforces a confirmation gate before it runs. Nothing executes silently. Recovery paths and policy basis surfaced at proposal time.

03 Every response is produced the same way

Two questions phrased differently get different answers, but they get answers produced by the same process. An auditor can verify the process. This is what regulated buyers actually need.

What it looks like

A governed response, annotated.

This is what a Raptor-governed response looks like in production. Every segment is tagged with a trust boundary and its evidential basis. The full response carries a provenance chain linking it to the execution that produced it. Actions require explicit confirmation before execution.

User query

What's the current authorization status for Project Sentinel, and can you extend the Phase 2 deadline to March 15?

Raptor-governed response
CONFIRMED basis: cryptographic_verification

Project Sentinel (ID: PRJ-2847) is currently authorized through Phase 2 with an active status. Authorization was granted on 2026-01-12 by ISSM J. Torres.

RETRIEVED basis: file_content

The current Phase 2 deadline is February 28, 2026. Three deliverables remain open against this phase.

EXECUTED Confirmation required

Proposed action: Extend Phase 2 deadline from 2026-02-28 to 2026-03-15 for Project Sentinel.

Confidence: RECOMMENDED — good confidence, solid evidence.

Basis: Deadline extensions require project-lead or higher authorization. Current actor has project-lead grant.

Risk if wrong: Downstream milestone dependencies may shift. Three open deliverables inherit new deadline.

Approve Reject
INFERRED basis: model_response

Based on current velocity, the three open deliverables are likely completable within the extended window. This assessment is probabilistic and not verified against resource allocations.

Response provenance ↓
execution_id: 7f3a2c91-...
intent_id: a8b4e1d0-...
input_hash: sha256:e3b0c44...
correlation_id: 9c1d4f82-...

Every governed response carries cryptographic provenance — verifiable, replayable, immutable.

Every segment is produced by the same deterministic process regardless of how the question is phrased. The CONFIRMED segment can't appear without cryptographic verification. The EXECUTED action can't run without passing the confirmation gate. The INFERRED segment is explicitly marked as model reasoning. An auditor sees exactly what the system verified, what it did, and what it guessed. Response structure simplified for illustration. Full schema in architecture docs.
Verified state

What's actually built.

4
AI providers, unified
32
REST API routes
11
Immutable Postgres tables
100%
Llama eval accuracy
  • 4 production-class AI providers behind a unified interface — Anthropic, OpenAI, Gemini, Together AI
  • OpenAPI spec and generated TypeScript SDK with trust-boundary types preserved
  • 9-unit UI design system implemented in Preact and deployed
  • 11 immutable Postgres tables with database-level append-only enforcement — every governed response is recoverable for audit
  • Cross-model eval: Llama 3.3 70B at 100% accuracy vs Claude Sonnet 4 at 97.8% — open-weight bridge empirically validated
  • MCP server functional with token auth (OAuth 2.1 designed)
  • SDVOSB certified for defense and government set-aside contracting
What's not built: self-hosted deployment, MCP registry publication. What is: pricing is locked, 14-day Workspace trial is live, pay-as-you-go Infrastructure access is available. We're early. The substrate is real.
The category

Deterministic-by-default isn't a feature. It's a category.

Regulated industries already have audit requirements that probabilistic-everywhere AI can't meet — SR 11-7 in finance, HIPAA in healthcare, IL4/IL5 in defense, the EU AI Act broadly. Public failures of ungoverned AI in regulated production are visible across industries. Adjacent solutions tell you what happened or test capability — they don't govern what's permitted to happen.

The deterministic-by-default position is open. Whoever ships the first credible substrate that wins a regulated buyer cohort defines the language the category uses.

Go deeper

Five audiences. Different sections.

Running AI in production?

If you've had an incident you couldn't explain to a customer or auditor, you're who Raptor is built for.

Read the buyer's view

Building AI products?

Three integration patterns, four model providers, an SDK and an MCP server. Read the code, then decide.

Read the developer's view

Model provider, integration partner, or channel?

Raptor's substrate is multi-provider by design. The right partnership shape depends on what you bring.

Read the partner's view

Evaluating substrate plays?

Self-funded, pre-revenue, solo founder, fully built substrate. Honest about all four. Read what we're asking you to evaluate.

Read the investor's view
Next step

Start with a conversation. Not a procurement track.

The first conversation is technical, not sales — you'll talk to the founder, not a sales engineer. Start with a 14-day Workspace trial, or tell us what you're building and we'll show you how Raptor governs it.

14-day Workspace trial · pay-as-you-go Infrastructure at $0.012/response · pricing details