tutorials
Intended Documentation
Platform Roadmap Programs
Public-facing overview of Intended post-120-day programs — capability truth, promise boundaries, and persona-specific value for each program.
Platform Roadmap Programs#
This page documents the six strategic programs that were closed out from the post-120-day roadmap. Each section describes delivered capability truth and explicit promise boundaries.
Info
This document uses Intended's capability-truth classification with production delivery focus: implemented (code exists and is deployable) and external dependency (outside the codebase boundary).
1. Multi-Step Workflow Engine#
What It Does#
Governs chains of AI agent actions as a single authorized workflow — not just individual intents. Each step in the chain is independently authorized, auditable, and fail-closed.
Current Capability Truth#
| Component | Status |
|---|---|
| Agent chain governance (7-state machine) | Implemented |
| Workflow database schema | Implemented |
| Workflow API routes | Implemented |
| Dynamic workflow planner | Implemented |
| Step-level authority evaluation | Implemented |
| Workflow rollback/compensation | Implemented |
What This Means for You#
Operator: You can define, schedule, trigger, and intervene in workflows with per-step authority checks and audit lineage.
Developer: The workflow API is live with advanced contracts for scheduling, triggers, analytics, and intervention.
Enterprise admin: Agent chain governance is fully implemented — multi-agent state tracking, escalation, and replay work today.
Promise Boundary#
Multi-step workflows will not execute steps without per-step authorization. We will not ship a workflow engine that bypasses the authority gate for convenience or performance.
2. Billing & Commerce#
What It Does#
Usage-based billing for Intended platform consumption — metering intent evaluations, managing subscriptions, and enforcing plan entitlements.
Current Capability Truth#
| Component | Status |
|---|---|
| Stripe connector (invoice, refund, list) | Implemented |
| Billing admin role | Implemented |
| Usage metering | Implemented |
| Subscription lifecycle | Implemented |
| Plan entitlements | Implemented |
What This Means for You#
Enterprise admin: Billing infrastructure is live with subscription lifecycle handling, usage metering, reconciliation visibility, and entitlement guardrails.
Developer: No billing-related API changes are needed in your integration. Entitlement enforcement will be transparent at the gateway level.
Promise Boundary#
Billing enforcement will be fail-closed for entitlement checks — if billing status is unknown, access is denied. We will not silently allow overages. Metering will be reconcilable against the audit trail.
3. Connector Expansion#
What It Does#
Extends Intended authorization enforcement to downstream systems — cloud providers, ITSM tools, CRM, observability platforms — through the connector SDK.
Current Capability Truth#
| Component | Status |
|---|---|
| Connector SDK framework | Implemented |
| Adapter registry & discovery | Implemented |
| GitHub Actions connector | Implemented |
| GitHub PR connector | Implemented |
| Slack connector | Implemented |
| Stripe connector | Implemented |
| Jira connector | Implemented |
| ServiceNow connector | Implemented |
| AWS, Azure, Salesforce, Datadog | Implemented |
What This Means for You#
Developer: The connector SDK is production-ready for custom connector development and conformance validation.
Operator: 10+ provider-native connectors are available with fail-closed capability enforcement.
Enterprise admin: Connector security model will enforce credential isolation and fail-closed behavior for all connectors.
Promise Boundary#
Every connector validates a decision token before execution. No connector operates without authorization.
4. SOC2 Type I/II Execution#
What It Does#
Achieves SOC2 Type I and Type II attestation, proving that Intended meets enterprise security and compliance standards.
Current Capability Truth#
| Component | Status |
|---|---|
| Immutable audit event ledger | Implemented |
| Cryptographic evidence chains | Implemented |
| Compliance pack export | Implemented |
| SOC2 policy pack lifecycle | Implemented |
| CC1–CC9 control mapping | Implemented |
| Continuous compliance monitoring | Implemented |
| Type I/II execution handoff | Implemented |
| External auditor-issued attestation document | External dependency |
What This Means for You#
Auditor: Evidence generation, integrity verification, engagement workflows, and attestation handoff verification are live.
Enterprise admin: Intended's audit infrastructure is production-ready. SOC2 attestation is a program milestone, not a platform feature — it validates the controls already in place.
Security engineer: Fail-closed controls, enforcement lineage, and operational readiness checks are all implemented and documented.
Promise Boundary#
We do not fabricate external certifications. Auditor-issued attestation documents remain external artifacts and are ingested/verified by the platform.
5. Cross-Tenant Benchmarking#
What It Does#
Provides anonymized performance benchmarks so operators can compare their authorization performance against industry peers.
Current Capability Truth#
| Component | Status |
|---|---|
| Per-tenant decision metrics (Prometheus) | Implemented |
| HTTP request metrics | Implemented |
| Policy simulation | Implemented |
| Cross-tenant aggregation pipeline | Implemented |
| Benchmark API | Implemented |
| Benchmark dashboard | Implemented |
What This Means for You#
Operator: Cross-tenant benchmark outputs are opt-in, anonymized, and confidence-scored.
Enterprise admin: Benchmark participation requires explicit tenant-level consent. No tenant data will be included in benchmarks without opt-in.
Promise Boundary#
Benchmarking will never expose tenant-identifiable data. If a cohort has fewer than 10 tenants, no benchmark data will be published for that cohort. Benchmarking is informational — it will not affect authorization decisions.
6. Intent Compiler & Domain LIM Ecosystem#
What It Does#
Makes Intended's Intent Compiler available as a standalone compile API and creates a marketplace for domain LIM packs.
Current Capability Truth#
| Component | Status |
|---|---|
| Intent Compiler engine | Implemented |
| LLM backends (Claude, OpenAI) | Implemented |
| Rules-based classification | Implemented |
| Policy pack framework | Implemented |
| Marketplace database schema | Implemented |
| Universal intent schema spec | Implemented |
| Marketplace API | Implemented |
| Federation protocol | Implemented |
| Ecosystem SDK | Implemented |
What This Means for You#
Developer: Intent Compiler is embedded in tenant evaluation pipelines and exposed through versioned service APIs.
Operator: Policy packs are available through marketplace workflows with provenance and compatibility enforcement.
Enterprise admin: Marketplace packs will be sandboxed and provenance-verified before installation. No pack can bypass authorization or exfiltrate data.
Promise Boundary#
Intent Compiler services enforce tenant isolation and versioned contracts. The marketplace does not permit unverified packs.
Related Resources#
- Capability Truth Matrix — current platform feature status
- Platform JTBD — persona workflow documentation
- Go-Live Runbook — production deployment checklist
- Architecture Overview — system components