IIP / Learning preview available / Production v1 pending

A clear release boundary.

The public learning preview is available now. Explore IIP locally with synthetic data, while keeping the production v1 release boundary clear.

PUBLIC SOURCE-ONLY PRERELEASE

learning-v0.84.0

Original project material is licensed under Apache-2.0; dependencies keep their own licenses. Docker builds the application from source and runs a disposable local lesson. No AWS account, provider keys, Kubernetes cluster, or host Python installation is needed.

Get the learning releaseFollow the walkthrough

The release includes a source archive and SHA256SUMS. Checksums verify download integrity, not an organizational signature. First launch needs Internet access to download dependencies and upstream images; no signed IIP container images or published SDK packages are included.

Read the versioned notes and limits →

What the lesson demonstrates

Eight local containers carry synthetic Bedrock/OpenAI-shaped OTel metadata through usage, calculated cost, app/team attribution, coverage gaps, and rule-based opportunities into the console and Grafana. You can also add a synthetic resource and follow a read-only investigation to its evidence citations. The sample investigator and cost rules do not invoke an LLM.

These are fixture prices and simulated requests—not live provider calls, actual bills, realized savings, or a production root-cause diagnosis.

No public v1 release yet

The learning preview uses known demo credentials, anonymous Grafana, internal plaintext connections, and disposable data. Keep it local. Do not connect customer workloads, load real secrets, or expose it to the Internet. It is not a supported production installation or an upgrade path to the separate persistent community preview.

Explore the public source →

What remains before v1

  1. 01

    Production distribution

    The source is public under Apache-2.0. Signed runtime artifacts, a supported-version policy, and verified public-artifact installation remain separate release work.

  2. 02

    Installation and first value

    Verify a fresh-machine install from published artifacts, clear prerequisites, supported profiles, and useful empty and error states.

  3. 03

    Real AI economics

    Qualify one live Bedrock invocation through OTel, the usage ledger, reviewed pricing, attribution, and Grafana. Keep estimates distinct from invoices.

  4. 04

    Operational lifecycle

    Complete supported backup/restore, upgrade/rollback, certificate lifecycle, retention, monitoring, and failure/recovery exercises.

  5. 05

    Compatibility and release evidence

    Define the supported feature matrix and contract stability policy; regenerate qualification evidence for the exact release candidate.

No release date or completion percentage is promised before the supported scope and evidence are settled. Broader providers, autonomous FinOps agents, and enterprise or regional availability remain later work.

Understand the current architecture →